XJENZAVOLUME 3 ISSUE 1, 2015

The Journal of the Malta Chamber of Scientists

www.xjenza.com

ONLINE

Editor-in-Chief: Giuseppe Di Giovanni

MALTA NEUROSCIECE NETWORK PROGRAMME

Xjenza Online - Journal of Malta Chamber of Scientistshttp://www.xjenza.org

Editorial Board

Editor-in-Chief

Prof Giuseppe Di GiovanniDepartment of Physiology and Biochemistry,University of Malta,Msida, Malta.MSD 2080Tel.: +356 2340 2776Fax: +356 2131 0577giuseppe.digiovanni@um.edu.mt

Associate Editors

Biomedical SciencesMario Valentinovalentino.mario@um.edu.mt

Cognitive and Social SciencesIan Thorntonian.thornton@um.edu.mt

EconomicsIan Cassarian.p.cassar@um.edu.mt

Engineering SciencePhilip Farrugiaphilip.farrugia@um.edu.mt

GeosciencesSebastiano D’Amicosebastiano.damico@um.edu.mt

Information and Communication TechnologiesNicholas Sammutnicholas.sammut@um.edu.mt

Medical SciencesJoseph Galeajosephgalea@yahoo.com

Life SciencesSandro Lanfrancosandro.lanfranco@um.edu.mt

David Mifsuddavid.a.mifsud@um.edu.mt

Social SciencesGodfrey Baldacchinogodfrey.baldacchino@um.edu.mt

Mathematical and Statistical ScienceLiberato Camilleriliberato.camilleri@um.edu.mt

Physics and Chemical SciencesDavid Magridavid.magri@um.edu.mt

Advisory Board Members

Prof Angela A. Xuereb Anastasi, University of Malta angela.a.xuereb@um.edu.mtProf David Eisner, Manchester University, UK eisner@manchester.ac.ukProf Frank Vella, University of Saskatchewan, Canada f.vella@sasktel.netProf Vincenzo Crunelli, Cardiff University, UK crunelli@cardiff.ac.uk

Project editor Copy Editor Editorial Assistant Web Administrator

Jackson Levi Said William Hicklin Caitlin Davies John GabarrettaDepartment of Physics, Department of Physics, School of Biosciences, Department of Chemistry,University of Malta, University of Malta, University of Cardiff, UK University of Malta,Msida MSD 2080, Malta. Msida MSD 2080, Malta. Life Sciences Building Msida MSD 2080, Malta.jsaid01@um.edu.mt whick01@um.edu.mt Museum Avenue, CF10AX john.gabarretta.09@um.

caitlin.davies@live.co.uk edu.mt

ii

Scope of JournalXjenza is the Journal of the Malta Chamber of Scientists and ispublished in an electronic format. Xjenza is a peer-reviewed, openaccess international journal. The scope of the journal encompassesresearch articles, original research reports, reviews, short com-munications and scientific commentaries in the fields of: mathe-matics, statistics, geology, engineering, computer science, socialsciences, natural and earth sciences, technological sciences, lin-guistics, industrial, nanotechnology, biology, chemistry, physics,zoology, medical studies, electronics and all other applied andtheoretical aspect of science.

The first issue of the journal was published in 1996 and the last(No. 12) in 2007. The new editorial board has been formed withinternationally recognised scientists, we are planning to restartpublication of Xjenza, with two issues being produced every year.One of the aims of Xjenza, besides highlighting the exciting re-search being performed nationally and internationally by Maltesescholars, is to provide insight to a wide scope of potential authors,including students and young researchers, into scientific publish-ing in a peer-reviewed environment.

Instructions for AuthorsXjenza is the journal of the Malta Chamber of Scientists and ispublished by the Chamber in electronic format on the website:http://www.mcs.org.mt/index.php/xjenza. Xjenza will considermanuscripts for publication on a wide variety of scientific topicsin the following categories

1. Communications2. Research Articles3. Research Reports4. Reviews5. Notes6. News7. Autobiography

Communications are short peer-reviewed research articles(limited to three journal pages) that describe new important re-sults meriting urgent publication. These are often followed by afull Research Article.

Research Articles form the main category of scientific pa-pers submitted to Xjenza. The same standards of scientific con-tent and quality that applies to Communications also apply toResearch Articles.

Research Reports are extended reports describing re-search of interest to a wide scientific audience characteristic ofXjenza. Please contact the editor to discuss the suitability oftopics for Research Reports.

Review Articles describe work of interest to the wide read-ership characteristic of Xjenza. They should provide an in-depthunderstanding of significant topics in the sciences and a criticaldiscussion of the existing state of knowledge on a topic based onprimary literature. Review Articles should not normally exceed6000 words.Authors are strongly advised to contact the EditorialBoard before writing a Review.

Notes are fully referenced, peer-reviewed short articles limitedto three journal pages that describe new theories, concepts anddevelopments made by the authors in any branch of science andtechnology. Notes need not contain results from experimental orsimulation work.

News: The News section provides a space for articles up tothree journal pages in length describing leading developments inany field of science and technology or for reporting items such asconference reports. The Editor reserves the right to modify orreject articles for consideration as ‘news items’.

Errata: Xjenza also publishes errata, in which authors correctsignificant errors of substance in their published manuscripts. Thetitle should read: Erratum: “Original title” by ***, Xjenza, vol.*** (year). Errata should be short and consistent for clarity.

Invited Articles and Special Issues: Xjenza regu-larly publishes Invited Articles and Special Issues that consist ofarticles written on invitation by the Editor or member of the edi-torial board.

Submission of ManuscriptsManuscripts should be sent according to the guidelines given here-after to submissionxjenzaonline@gmail.com.

Referees All manuscripts submitted to Xjenza are peer re-viewed. Authors are requested to submit with their manuscriptthe names and addresses of three referees, preferably from over-seas. Every effort will be made to use the recommended reviewers;however the editor reserves the right to also consult other compe-tent reviewers.

Conflict of Interest Authors are expected to disclose anycommercial or other associations that could pose a conflict of in-terest in connection with the submitted manuscript. All fundingsources supporting the work, and institutional or corporate affil-iations of the authors, should be acknowledged on the title pageor at the end of the article.

Policy and Ethics The work described in the submittedmanuscript must have been carried out in accordance with TheCode of Ethics of the World Medical Association (Declaration ofHelsinki) for experiments involving humans (http://www.wma.net/en/30publications/10policies/b3/index.html); EU Direc-tive 2010/63/EU for animal experiments (http://ec.europa.eu/environment/chemicals/lab_animals/legislation_en.htm);Uniform Requirements for manuscripts submitted to Biomedicaljournals (http://www.icmje.org). This must be stated at anappropriate point in the article.

Submission, Declaration and Verification Submis-sion of a manuscript implies that the work described has not beenpublished previously (except in the form of an abstract or as partof a published lecture or academic thesis), that it is not underconsideration for publication elsewhere, that it has been approvedfor publication by all authors, and tacitly or explicitly, by the re-sponsible authorities where the work was carried out, and that, ifaccepted, it will not be published elsewhere in the same form, inEnglish or in any other language, including electronically, withoutthe written consent of the copyright-holder.

Permissions It is the responsibility of the corresponding au-thor of a manuscript to ensure that there is no infringement ofcopyright when submitting material to Xjenza. In particular,when material is copied from other sources, a written statement isrequired from both the author and/or publisher giving permissionfor reproduction. Manuscripts in press, unpublished data andpersonal communications are discouraged; however, correspond-ing authors are expected to obtain permission in writing from atleast one author of such materials.

Preparation of ManuscriptsXjenza accepts submissions in MS Word, Libre Office Writerand LATEX with the latter being the preferred option. Anyonesubmitting in LATEX should use the journal template, the latestversion of which can be found at http://github.com/hicklin/

Xjenza-Journal-Template. All the necessary files to run theLATEX document should be supplied together with the renderedPDF.

iii

If a word processor is used the styling should be kept to aminimum, only introducing bold face, italics, subscript and su-perscript text where the context requires it. Text should be insingle-column format and the word processor options should notbe used in order to justify text or hyphenate words. Togetherwith the native format of the word processor, a pdf, generated bythe word processor, must be given. Furthermore, artwork shouldbe in accordance to the artwork guidelines give below and mustbe submitted separately from the word processor file. Similarly,the bibliographic data of the cited material should be submittedseparately as an Endnote (*.xml), Research Information Systems(*.ris), Zotero Library (zotero.splite) or a BiBTEX (*.bib) file.

Article Structure

A manuscript for publication in Xjenza will ordinarily consistof the following: Title page with contact information, Abstract,Highlights, Keywords, Abbreviations, Introduction, Materials andMethods, Results, Discussion, Conclusion, Appendices and Refer-ences.

The manuscript will be divided into clearly defined numberedsections. Each numbered subsection should be given a brief head-ing. Each heading should appear on its own separate line. Subsec-tions should be used as much as possible when cross-referencingtext: refer to the subsection by the section number as opposed tosimply ‘the text’.

Title page• Title should be concise yet informative. Titles are often

used in information-retrieval systems. Avoid abbreviationsand formulae where possible.

• Author names and affiliations. Present the authors’ affilia-tion addresses (where the actual work was done) below thenames. Indicate all affiliations with a lower-case superscriptnumber immediately after each author’s name and in front ofthe appropriate address. Provide full postal address of eachaffiliation, including the country name and, if available, thee-mail address.

• Corresponding author. Clearly indicate who will handlecorrespondence at all stages of refereeing and publication,including post-publication. Ensure that telephone and faxnumbers (with country and area code) are provided in ad-dition to the e-mail address and complete postal address.Contact details must be kept up to date by the correspond-ing author.

• Present/permanent address. If an author has changed theaddress since the work described, this can be indicated asa footnote to the author’s name. The address at which theauthor actually did the work must be retained as the main,affiliation address. Superscript Arabic numerals are used forsuch footnotes.

Abstract A concise and factual abstract is required of up toabout 250 words. The abstract should state briefly the back-ground and purpose of the research, the principal results and ma-jor conclusions. An abstract is often presented separately fromthe article, so it must be able to stand alone. For this reason,references and non-standard abbreviations should be avoided. Ifessential, these must be defined at first mention in the abstractitself.

Abbreviations Define abbreviations that are not standardin this field in a footnote to be placed on the first page of thearticle. Such abbreviations that are unavoidable in the abstractmust be defined at their first mention as well as in the footnoteand should be used consistenly throughout the text.

Introduction State the objectives of the work and providean adequate background, avoid a detailed literature survey or asummary of the results.

Material and Methods Provide sufficient detail to allowthe work to be reproduced. Methods already published should beindicated by a reference: only relevant modifications should bedescribed.

Results Results should be clear and concise. Num-bered/tabulated information and/or figures should also be in-cluded.

Discussion This should explore the significance of the resultsof the work, yet not repeat them. Avoid extensive citations anddiscussion of published literature. A combined section of Resultsand Discussion is often appropriate.

Conclusion The main conclusions based on results of thestudy may be presented in a short Conclusions section. This maystand alone or form a subsection of a Discussion or Results andDiscussion section.

Appendices Formulae and equations in appendices shouldbe given separate numbering: Eq. (A.1), Eq. (A.2), etc.; in asubsequent appendix, Eq. (B.1) and so on. Similarly for tablesand figures: Table A.1; Fig. A.1, etc.

Acknowledgements Collate acknowledgements in a sepa-rate section at the end of the article before the references and donot, therefore, include them on the title page, as a footnote to thetitle or otherwise. List here those individuals who provided assis-tance during the research (e.g., providing language help, writingassistance or proof reading the article, etc.).

Units Follow internationally accepted rules and conventions:use the international system of units (SI). If other units are men-tioned, please give their equivalent in SI. Anyone using LATEXshould use the package siunitx in all cases.

Footnotes Footnotes should be used sparingly. Numberthem consecutively throughout the article, using superscript Ara-bic numbers. Many word processors build footnotes into the text,and this feature may be used. Should this not be the case, indi-cate the position of footnotes by a superscript number in the textand present the footnotes themselves separately at the end of thearticle. Do not include footnotes in the Reference list.

Table Footnotes Indicate each footnote in a table with asuperscript lower case letter.

Artwork Electronic artwork General points:• Make sure you use uniform lettering and sizing of your orig-

inal artwork.• Save text in illustrations as ‘graphics’ or enclose the font.• Only use the following fonts in your illustrations: Arial,

Courier, Times, Symbol or Computer Modern Roman, thelatter is preferred.

• Number the illustrations according to their sequence in thetext.

• Name your artwork files as ‘figx’ or ‘tabx’ where x corre-sponds to the sequence number in your document.

• Provide captions to illustrations separately.• Produce images near to the desired size of the printed version

or grater.• Make sure that the artwork has no margins and borders.• Submit each figure as a separate file.

Formats Regardless of the application used, when your elec-tronic artwork is finalised its file format should be one of thefollowing (note the resolution requirements for line drawings,halftones, and line/halftone combinations given below):

• PDF or SVG: Vector drawings. Embed the font or save thetext as ‘graphics’.

• JPEG or PNG: Color or grayscale photographs (halftones):always use a minimum of 300 dpi.

• JPEG or PNG: Bitmapped line drawings: use a minimum of1000 dpi.

iv

• JPEG or PNG: Combinations bitmapped line/half-tone(color or grayscale): a minimum of 500 dpi is required.

Where possible use a vector format for your artwork (PDF orSVG). If this is not possible, supply files that have and adequateresolution.

Colour Artwork Make sure that color artwork files are inan acceptable format (JPEG, PNG, PDF or SVG) and have thecorrect resolution.

Figure Captions Ensure that each illustration has a cap-tion. Supply captions separately, not attached to the figure. Acaption should comprise a brief title (not on the figure itself) and adescription of the illustration. Keep text in the illustrations them-selves to a minimum, but explain all symbols and abbreviationsused.

Tables Number tables consecutively in accordance with theirappearance in the text. Place footnotes to tables below the tablebody and indicate them with superscript lowercase letters. Avoidvertical rules. Be sparing in the use of tables and ensure thatthe data presented in tables do not duplicate results describedelsewhere in the article. Large tables should be submitted in CSVformat.

Citations and References Reference and citation stylesfor manuscripts submitted to Xjenza should be in accordance tothe APA v6 style.

Citation in text References to cited literature in thetext should be given in the form of an author’s surname and theyear of publication of the paper with the addition of a letter forreferences to several publications of the author in the same year.For further information regarding multiple authors consult theAPA v6 guidelines. Citations may be made directly

Kramer et al. (2010) have recently shown . . .

or parenthetically

as demonstrated (Allan, 2000a, 2000b, 1999; Allan and Jones,1999).

Groups of references should be listed first alphabetically, thenchronologically. When writing in LATEX use \textcite and\parencite for the respective cases mentioned.

The reference section Every reference cited in the textshould also be present in the reference list (and vice versa). Thereference list should also be supplied as an Endnote (*.xml), Re-search Information Systems (*.ris), Zotero Library (zotero.splite)or a BiBTEX (*.bib) file. Unpublished results and personal com-munications are not recommended in the reference list, but maybe mentioned in the text. If these references are included in thereference list they should follow the standard reference style ofthe journal and should include a substitution of the publicationdate with either ‘Unpublished results’ or ‘Personal communica-tion’. Citation of a reference as ‘in press’ implies that the itemhas been accepted for publication.

References should be arranged first alphabetically and then fur-ther sorted chronologically if necessary. More than one referencefrom the same author(s) in the same year must be identified bythe letters ’a’, ’b’, ’c’, etc., placed after the year of publication.Consult the APA v6 guidelines for multiple authors. Below aresome examples of referencing different bibliographic material.

Reference to a Journal Publication:

Agree, E. M. and Freedman, V. A. (2011). A Quality-of-LifeScale for Assistive Technology: Results of a Pilot Study ofAging and Technology. Phys. Ther., 91(12):1780–1788.

McCreadie, C. and Tinker, A. (2005). The acceptability ofassistive technology to older people. Ageing Soc., 25(1):91–110.

Reference to a Book:

Brownsell, B. (2003). Assistive Technology and Telecare: Forg-ing Solutions for Independent Living.Policy Press, Bristol.

Fisk, M. J. (2003). Social Alarms to Telecare: Older People’sServices in Transition.Policy Press, Bristol, 1st edition.

Reference to a Chapter in an Edited Book:

Brownsell, S. and Bradley, D. (2003). New Generations of Tele-care Equipment. In Assist. Technol. Telecare Forg. Solut.Indep. Living, pages 39–50.

Web references The full URL should be given togetherwith the date the reference was last accessed. Any further in-formation, if known (DOI, author names, dates, reference to asource publication, etc.), should also be given. Web referencescan be listed separately or can be included in the reference list.

References in a Special Issue Please ensure that thewords ‘this issue’ are added to any references in the list (and anycitations in the text) to other articles in the same Special Issue.

Journal Abbreviations Journal names should be ab-breviated according to:

-Index Medicus journal abbreviations: http://www.nlm.nih.

gov/tsd/serials/lji.html;-List of title word abbreviations: http://www.issn.org/

2-22661-LTWA-online.php;-CAS (Chemical Abstracts Service): http://www.cas.org/

sent.html.

Video data Xjenza accepts video material and animation se-quences to support and enhance the presentation of the scientificresearch. Authors who have video or animation files that theywish to submit with their article should send them as a separatefile. Reference to the video material should be clearly made intext. This will the modified into a linked to the paper’s supple-mentary information page. All submitted files should be properlylabelled so that they directly relate to the video files content. Thisshould be within a maximum size of 50 MB.

Submission check listThe following list will be useful during the final checking of amanuscript prior to sending it to the journal for review. Pleaseconsult this Guide for Authors for further details of any item.

• One author has been designated as the corresponding authorwith contact details:

– E-mail address.– Full postal address.– Telephone and fax numbers.

• All necessary files have been sent, and contain:– All figures are given separately in PDF, SVG, JPEG

of PNG format.– Caption for figures is included at the end of the text.– All tables (including title, description, footnotes) are

included in the text and large tables have been givenseparately as CSV.

– The reference list has been given in XML, RIS,zotero.splite or BIB file format.

• Further considerations– Abstract does not exceed about 250 words.– Manuscript has been ‘spell-checked’ and ‘grammar-

checked’.– References are in the required format.– All references mentioned in the reference list are cited

in the text, and vice versa.– Bibliographic data for all cited material has been

given.– Permission has been obtained for use of copyrighted

material from other sources (including the Web).

v

– A PDF document generated from the word processorused is given.

After AcceptanceUse of the Digital Object Identifier The Digital Ob-ject Identifier (DOI) may be used to cite and link to electronicdocuments. The DOI consists of a unique alpha-numeric char-acter string which is assigned to a document by the publisherupon the initial electronic publication. The assigned DOI neverchanges. Therefore, it is an ideal medium for citing a document,particularly ‘Articles in press’ because they have not yet receivedtheir full bibliographic information. When you use a DOI to cre-

ate links to documents on the web, the DOIs are guaranteed neverto change.

Proofs, Reprints and Copyright Authors will nor-mally be sent page proofs by e-mail or fax where available. Alist of any necessary corrections should be sent by fax or email tothe corresponding editor within a week of proof receipt to avoidunnecessary delays in the publication of the article. Alterations,other than essential corrections to the text of the article, shouldnot be made at this stage. Manuscripts are accepted for publica-tion on the understanding that exclusive copyright is assigned toXjenza. However, this does not limit the freedom of the author(s)to use material in the articles in any other published works.

Mifsud, J. (2015).Xjenza Online, 3:1–5.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.org

Special Issue Introductory Article

Malta’s participation in Europe’s oldest research cooperation frame-work

J. MifsudCOST National Contact Point, COST Malta Representative on the Committee of Scientific Officers, Malta Councilfor Science and Technology, Kalkara

Abstract. COST (European Cooperation in Scienceand Technology) is the longest-running European re-search framework supporting networking and knowledgesharing amongst science and technology communities inEurope. It does not fund research as such but greatlyfacilitates the networking and the coordination betweenresearchers working on nationally funded research acrossEurope. The wide range of networking activities suppor-ted by COST, such as meetings, workshops, short termscientific missions, publications and training schools andeasy administration made it even very amenable to re-searchers based in Malta. Since 2011 Malta has par-ticipated in a constantly growing number of runningCOST Actions and this has resulted in better exposureopportunities for them. More importantly, early stageresearchers have been given an invaluable opportunity toincrease their contacts circles in their fields of interest.This has resulted in unique opportunities for participa-tion and networking in research for researchers based inMalta on a European scale.

1 Introduction

This issue of Xjenza may seem somewhat eclectic, in-cluding many diverse and unrelated topics. Many maywonder what does research concerning supramolecularchemistry in water has to do with drug resistance incancer stem cells, soil erosion and 3D video coding?Moreover, are studies on the symbiotics of history andsocial psychology, and research on accessibility as an in-dicator of transport equity considered as science?

The brief answer to these questions is COST. Thisspecial issue of Xjenza highlights just some of the re-search that has resulted from the nearly 200 COST ac-tions, researchers based in Malta are or have particip-

ated in. You others may wonder what exactly COST is?How can I get involved? How will it benefit myself andthe research landscape in Malta?

2 What is COST?

COST (European Cooperation in Science and Tech-nology) (www.cost.eu) is in fact the longest-runningEuropean research framework supporting networkingand knowledge sharing amongst science and technologycommunities in Europe (Halen, 2014). It does not fundresearch as such but greatly facilitates the networkingand the coordination between researchers working onnationally funded research across Europe. Thus it con-tributes greatly to narrowing the gap between science,policy makers and society. Co-operation in COSTis inclusive and open, fosters new and excellentideas through the sharing of knowledge (see Figure1).

Since its establishment in 1971, COST has been doingthis through supporting networks (called COST Ac-tions) co-ordinated by Management Committee mem-bers (MCs). A huge advantage of these MCs is thateach COST country is allowed to nominate up to twoMCs per action irrespective of size, thus ensuring in-clusiveness and widening. This is a big advantage forMalta. As such, COST has always been the corner-stone of the European research funding landscape sinceit opens up huge networks to smaller and more peri-pheral COST countries, such as Malta. Every year anaverage of 30 000 researchers benefit from COST fund-ing (Dietl, 2014).

Over the course of its 40 year history, COST has alsobeen able to constantly adapt itself to the changing en-vironment defining research policy in Europe. It hasa unique bottom up approach, using open calls with

Correspondence to: J. Mifsud (janet.mifsud@um.edu.mt)

c© 2015 Xjenza Online

Mifsud, J. (2015).Xjenza Online, 3:1–5. 2

Figure 1: Co-operation in COST is inclusive and open, fos-tering new and excellent ideas through the sharing of know-ledge.

no pre-defined priorities for research. It also promotesinterdisciplinary approaches, knowledge sharing, talentdevelopment and assists research communities in findingtheir own ways to tackle important societal challenges.

Since its inception, COST has provided a unique ap-proach that enables the sharing of concepts and sci-entific developments across 35 European countries andhas even attracted the participation of many non EUcountries.

There are three main features that make it a uniquetool:

Capacity COST acts to connect high-quality scientificminds both across Europe and internationally andenables the creation of research communities invarious fields.

Networking As a networking platform, it supports re-searchers to connect and build consortia that canlead to the submission of transnational project pro-posals for funding from the EU’s Framework Pro-gramme. This networking is interdisciplinary, facil-itating the participation from researchers outsidethe academic community, such as SMEs, public en-tities and NGOs.

Impact COST enables the formulation of publicationsand promotes the dissemination of information thatincreases impact on policy-makers and decision-making bodies (Armeni and Mifsud, 2015).

COST has also always been a frontrunner in bridging

the gap between policy, research and end users, by pro-moting connections between COST Actions and otherEuropean S&T policy concepts with relevant stake-holder communities. It has provided the basis for in-creased societal impact of research and S&T innovationin Europe and beyond.

3 Malta in COST

Malta first joined COST in 1996 before it became a fullEU member in 2004. The Malta Council for Scienceand Technology, as the managing authority for COST inMalta, then recognised the key role COST participationcould have in facilitating the entry of researchers basedin Malta to wider European Research networks andfunds such as in the Framework Programmes. Moreoverthe wide range of networking activities (tools) suppor-ted by COST, such as meetings, workshops, short termscientific missions, publications, and training schools,made it even more amenable to researchers based inMalta.

Since 2011 Malta has participated in a constantlygrowing number of running COST Actions. In 2014, theMaltese research community not only participated in theManagement Committee (MC) of 149 Actions (Table 1)but had access to all their networking activities (Halen,2015).

The number of Maltese participations to network-ing activities (meetings, workshops, STSM, TrainingSchools) has also been growing since 2011 (Table 1).In 2014, Maltese researchers participated in 200 COSTAction activities.

COST also promotes the participation of Early CareerInvestigators - ECI (less than PhD + 8 years) in all ofits networking activities (meetings, workshops, STSM,Training Schools). In 2014, almost half of the Malteseparticipations in networking activities (as shown in theprevious Table 1) were from Early Career Investigators-ECI (Table 1).

The Maltese research and innovation system benefitsmore and more from COST Actions’ budget. In fact in2014, around EUR 188,000 benefited Maltese research-ers (Table 1). This includes participations in meetings,workshops, STSM, Training Schools, Local OrganiserSupport, but also Maltese Grant Holder institutions’ ad-ministrative support – FSAC.

In 2014, Maltese researchers participated in 15 ShortTerm Scientific Missions (STSMs). Thirteen Maltesetrainees and one Maltese trainer were also involved intraining schools (Table 1).

Maltese institutions also gained increasing visibilitythanks to meetings, Training Schools and STSMs beinghosted in their premises. In 2014, 4 COST meetings, 1Training School and 5 STSMs were held in Malta help-ing to strengthen the existing networks and foster col-

www.xjenza.org

3 Mifsud, J. (2015).Xjenza Online, 3:1–5.

Year 2011 2012 2013 2014

COST actions with participation from Malta 47 70 116 149Maltese participation in COST networking activities 46 75 161 200Maltese early career investigators participation in COST networkingactivities

18 25 60 82

COST networking budget transferred to Malta (in euros) 42 233 72 674 141 883 188 015

COST Short Term Scientific Missions (STSM) and trainingschool participation from Malta

STSM participants 4 3 4 15Trainees in Training Schools 3 13 13Trainers in Training Schools 1 1 1

COST meetings, training schools, STSMs and Annual ProgressConference (APCs) organised in Malta

Meetings 1 1 3 4Training Schools – 1 2 1STSM 2 1 7 5APC – – 1 4

Table 1: COST action statistics (Halen, 2015).

laboration links between researchers, institutions andultimately countries. They also helped to increase thevisibility of hosting institutions. Maltese institutionsseem to be particularly attractive to researchers com-ing on STSMs (Table 1). In addition, Malta also hadthe honour to host five Annual Progress Conferences(APCs) from five of the ten COST domains in 2013 and2014.

4 COST impact in Malta

In 2014, MCST carried out a survey among Malta- basedCOST MC members in order to gauge the success, fail-ure and impact of COST since its initiation in Malta,aiming to identify the resulting impact of this participa-tion in COST on the local research arena and innovationinitiatives (Armeni & Mifsud, 2014).

5 Who is the Maltese COST re-searcher?

The survey was carried out by means of an online ques-tionnaire using Survey Monkey R©. There were a totalnumber of 100 responses, from 193 potential respondentsthat were participating in COST actions across Maltaand Gozo at the time the survey was taken. As alsoconfirmed by the data provided in Table 1, the majorityof respondents, 37.10 %, were between 30 and 39 yearsof age, followed by 25.81 % between 40 and 49 years ofage, 16.94 % between 50 and 59 years of age and 13.71 %between 21 and 29 years of age. Only 3.23 % noted thatthey were 60 years of age or older (Table 1). In this

respect COST has reached out to and involved com-paratively high numbers of young researchers below theage of 40 in the majority of its domains. Such capacitybuilding and opportunities for early stage researchers,who are key human resources, will thus contribute toa continued growth of innovation, product developmentand commercialisation in Malta.

In terms of career status, most respondents notedthat they hold the position of senior lecturers (27.42 %),whereas 19.35 % noted that they are lecturers, 9.68 %noted that they are associate professors and 3.23 %noted that they are professors. There was a positiveuptake from PhD students (13.71 %) and post-doctorateresearchers (7.26 %) indicating an increasing interest bythe younger population of researchers across the do-mains and the involvement and integration of a morejuvenile cohort of academics and scientists in the vari-ous research agendas of the different COST actions.

Significantly lower were percentages for participationfrom industry (1.61 %) and NGO’s (4.03 %). In addi-tion, there is very limited involvement reported fromthe public entities and the public sector suggesting thatmore efforts could be focused to generate awareness andinterest there. Academia continues to enjoy a stronghold in the programme. The lack of participation fromindustry could imply that COST is not yet understoodby Maltese industry as potentially being a significantcontributor towards their participation in transnationalconsortia or in research arenas. Increased efforts, dir-ected at this particular sector, could result in increasedparticipation when relevant domains/actions arise.

www.xjenza.org

Mifsud, J. (2015).Xjenza Online, 3:1–5. 4

6 What are the outcomes of par-ticipation in COST?

Most respondents stated that their COST action mem-bers intended to continue to co-operate once the COSTaction was completed (80 %). Nonetheless, some com-mented that any speculation of potential future ac-tion, at present, would be premature (29 %). An equalamount simply noted ’future collaborations’, such asthrough access to infrastructures and joint researchactivities, as possible sources of action (29 %) Othershighlighted future research contributions or submissionsunder Horizon 2020 as a possible way forward.

7 What can be done to improveMalta COST participation?

Participation in COST actions has had its benefits anddifficulties. The main benefits noted by the respondentswere: networking, opportunities for cooperation and col-laboration with European counterparts, increased ex-posure at a European level, opportunities for trainingand staff/student exchange and access to research infra-structures. Furthermore, the majority of respondentswho claimed prior participation in, or organisation of,COST events, both locally and abroad, noted that theoutput and impact was well worth the return on invest-ment, particularly in terms of time and effort.

The main obstacles were: financial and administrat-ive burdens, lack of time and resources, bureaucracy,language barriers, travel requirements and other work.In this respect, it was noted that the provision of morefrequent information sessions and assistance in sourcingand joining existing consortia, coupled with more handson guidance in the compilation and submission of re-search proposals for Horizon 2020 would be beneficial.This, as well as, funding to conduct further research re-mains a key concern for most respondents. Notably, lackof financial support implies lack of advancements in re-search and analytics in each domain which could placeMalta at a disadvantage when compared to other coun-tries. For this purpose, some respondents suggested apotential allocation of some resources that may act asan incentive for local researchers to construe researchgroups in areas that are deemed relevant to work upon.The areas may not have direct relevance to Malta, butMalta can act as a cost effective base where researchcan be carried out, generating income and knowledge.In addition, continued efforts towards increasing oppor-tunities for the mobility of researchers across Europe,to foster scientific excellence, is recommended togetherwith an alignment of the selection of local participa-tion in specific COST actions with the national scientificpriorities’ as outlined in the National Research and In-

novation Strategy 2020 (Malta Council for Science andTechnology, 2014).

In addition, the majority cohort of academic seniorlecturers as MC members and the lack of participationfrom industry, raises concerns, particularly, in view ofthe need to bridge the gap between research in aca-demia and industry. Lack of easily accessible inform-ation, easier access to academics, lack of available timeand resources and lack of tangible and/or immediatereturn on investment could be precluding players fromindustry from actively participating in COST. To thisend, one-to-one meetings, networking events, informa-tion sessions and the CNC’s or MCST staff participa-tion in events organized by other bodies representingindustry and young researchers, could be beneficial. Inaddition, in line with efforts to mediate gender equal-ity and balance in scientific spheres, it is advised thatmore emphasis is made on mediating the current genderbalance in COST actions

8 COST: the next steps

COST has been proud of its ability of renewal. Therenewal started in 2010, in response to the COST FP7Mid-Term Evaluation (Horvat et al., 2010) and throughthe COST Strategy (COST CSO, 2011) aims to achievebest performance, output orientation, better coopera-tion, and good governance. The new phase has beencharacterised by reviewing the institutional, operationaland administrative levels of the COST Framework andthe creation of a new independent legal entity, the‘COST Association’, tasked with becoming COST’simplementing agent in 2014 (COST CSO, 2013).

Over the last few years, COST has also adapted andchanged to respond to the requirements and challengesin Horizon 2020 ensuring that COST actions are fo-cussed on Spreading excellence and widening particip-ation as well as ensuring Europe in a changing world– inclusive, innovative and reflective societies (COSTCSO, 2014). Adaptations have always been made keep-ing in mind the above principles that govern the COSTFramework from both a strategic and operational pointof view: supporting excellence and being openand inclusive. The driving forces (deriving from theprinciples described above) of fostering new ideas,sharing knowledge and output orientation havebeen enabling COST to make significant contributionsto the competitiveness and overall development of theEuropean research landscape and assisting European re-search communities in overcoming the many challengesthey face and thereby contributing to Europe’s Innova-tion Union goals.

Researchers based in Malta are now integrated ineach part of the COST process, from participation inthe high level policy Committee of Scientific Officers

www.xjenza.org

5 Mifsud, J. (2015).Xjenza Online, 3:1–5.

(CSO), having a representative on the strategic Com-mittee of Scientific Experts, a full cohort of eighteenReview Panel Experts and also close to 300 COST ex-perts. Malta’s success in COST has often been high-lighted by the COST Association and Malta has evenbeen invited to be Vice-Chair for the COST workinggroup on Inclusiveness and Widening.

Malta’s participation and investment in COST hasresulted in better exposure opportunities for variousparties. More importantly, the younger cohort of re-searchers have been given an invaluable opportunity toincrease their contacts circles in their fields of interest.This, coupled with opportunities for participation in re-search on a European scale, opportunities for specialisa-tion through joint collaborations and opportunities toshowcase our local potential, continue to support andadvocate Malta’s necessity to participate in this pro-gramme and acts as an incentive for local participationto continue to be supported and encouraged.

References

Armeni, L. & Mifsud, J. (2014). COST: European Co-operation in Science and Technology. A Review forMalta. Policy, Programme Implementation, Pro-jects Unit, Malta Council for Science and Techno-logy. Kalkara, Malta.

COST CSO. (2011). COST Strategy. Renewing ourstrengths... shaping our future. COST Office.Brussels, Belgium.

COST CSO. (2013). COST Association Statues. COSTAssociation. Brussels, Belgium.

COST CSO. (2014). Action Plan of the Frame-work Partnership Agreement between the EuropeanCommission and the COST Association for imple-menting the COST activities under Horizon 2020.COST Association. Brussels, Belgium.

Dietl, M. (2014). COST at a turning point: A uniqueframework for pan-European S&T cooperation asclear demonstration of European values. COST Of-fice. Brussels, Belgium.

Halen, C. (2014). About COST. How to join a COSTaction. COST Office. Brussels, Belgium.

Halen, C. (2015). How can Maltese researchers andMaltese research and innovation system benefitfrom participation in COST? COST Association.Brussels, Belgium.

Horvat, M., Acheson, H., Demonte, V., Edler, J., Mus-tonen, R. & Vas, I. (2010). FP7 Mid-Term Evalu-ation of COST. COST Office. Brussels, Belgium.

Malta Council for Science and Technology. (2014). Na-tional Research and Innovation Strategy 2020.Malta Council for Science and Technology.Kalkara, Malta.

www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.01

Research Article

‘Everyone is a winner, help is just a push of a button away . . . ’: TheTelecare Plus service in Malta

M. FormosaUniversity of Malta, Gerontology Unit, Faculty for Social Wellbeing

Abstract. This paper reports on a research studyon the role of assistive technologies in later life. Re-search questions included what is the impact of assist-ive technologies on the quality of life of older service-users, and to what extent does assistive technology leadto an improved quality of life for subscribers and in-formal carers? The chosen method of enquiry was acase-study of the Telecare Plus service in Malta. A totalof 26 semi-structured interviews were held with a con-venience sample of 26 people aged 60-plus about theiruse and experience of this particular telecare system.The Telecare Plus service was found to contribute pos-itively to subscribers’ levels of emotional and physicalwellbeing, interpersonal relations and personal develop-ment, as well as towards the quality of life of informalcarers. However, research also highlighted a range ofchallenges that stood in the way of increased adoptionrates of the Telecare Plus service by older people. Thefact that the field of assistive technologies in Malta lacksefficient and clear business models constitutes anotherbarrier towards the take up of such services.

1 Introduction

All over the world, governments are facing the multi-dimensional consequences of an ageing population. Con-temporary social and political agendas are pepperedwith issues such as older people’s care, healthy ageing,and age-friendly housing - amongst others. However,concerns about the most cost-effective ways to providesocial and health care services are also being addressedkeenly. It is within such an emergent framework thatpolicies promoting assistive technology have emerged.From humble beginnings in the late 1940s, when somesheltered housing tenants in the United Kingdom hadthe facility to ring a bell in a warden’s home, nowadays

assistive technologies have revolutionised the delivery ofhealth and social care to older people, enabling them tolive healthier, safer and more independent lives.

Throughout this article, ’assistive technology’ isdefined as “any device or system that allows an indi-vidual to perform a task that they would otherwise beunable to do, or increases the ease and safety with whichthe task can be performed” (Cowan and Turner-Smith,cited in McCreadie and Tinker (2005)). In other words,assistive technology refers to any item, or piece of equip-ment, that is used to increase, maintain or improve thefunctional capabilities of individuals and independenceof people with cognitive, physical, or communicationdifficulties. Although the dominant stereotypical viewis that older persons are unable to master new techno-logical advances, and are apathetic to how technologycan assist them in achieving higher levels of active andsuccessful lifestyles, the available literature suggests thecontrary (Sixsmith & Gutman, 2013). Indeed, there isno shortage of literature on how assistive technologies -ranging from community alarms to fall equipment - havebeen mastered by relatively frail and vulnerable olderpeople to help them stay independent and in their homesfor as long as possible (Mann, 2005). Assistive technolo-gies also enable healthcare professionals to manage datain order to monitor various chronic conditions associatedwith old age, as well as to deliver services beyond theconfines of health and care institutions,whilst also facil-itating effective responses to critical events (Sixsmith &Gutman, 2013).

Various literature has recognised the role and contri-bution of assistive technology on the quality of lives ofolder persons (Horner, Soar & Krich, 2009). Few stud-ies, however, presented views or reactions of potentialusers of telecare services, and how this assistive techno-logy is used and experienced (Percival & Hanson, 2006).

Correspondence to: M. Formosa (marvin.formosa@um.edu.mt)

c© 2015 Xjenza Online

7 Formosa, M. (2015).Xjenza Online, 3:6–16.

The research reported herein addresses such a shortcom-ing by addressing questions such as: what is the impactof assistive technologies on the quality of life of olderservice-users? to what extent does assistive technologylead to improved quality of life, maintenance of abilityto remain at home, reduced burden placed on carers;and improved support for people with long-term healthconditions? The chosen method of enquiry was to under-take a case-study of the Telecare Plus service in Malta,by conducting semi-structured interviews through thetelephone with a convenience sample of 26 people aged60-plus that queried their use and experience of thisparticular telecare system1. Findings generated muchneeded knowledge and understanding about both thepotential benefits and possible contra-indications of theTelecare Plus service towards the quality of life of olderusers.

2 Assistive technologies andolder persons

In the realm of later life, assistive technology contrib-utes to the maintenance of human dignity, in the face ofthe fact that as people get older they often experiencedecreased financial capacity resulting from increasedfrailty and/or cognitive limitations. The scope of assist-ive technology in later life embodies a ‘social model’ ofdisability, which acknowledges that older persons’ dis-ability ascends from the interface between their phys-ical/mental capacities and the environment, especiallytheir living environments (McCreadie & Tinker, 2005).Assistive technologies offer assistance in the form of arange of equipment that boost the living environment,and thus, prolong older people’s ability to function com-mendably in the face of increasing dependency levels. Asa result, the application of assistive technologies aimsto (i) enable those who wish to age in their own res-idences by maintaining independence, extending capab-ility and productivity, and permitting more informedchoices; (ii) aid older people who reside in care facilit-ies and their formal/informal carers by facilitating com-munication patterns, and improving the monitoring andtreatment of conditions; (iii) support families in the pro-vision of the assistance and care that are needed by olderpersons, as well as to facilitate their roles as part of theholistic care team; and (iv), foster innovations that area response to ‘clinical, social, or personal need’ (Horneret al., 2009).

1The Maltese archipelago is a European Union Member State.It consists of three islands - Comino, Gozo and Malta, 93 kilo-metres south of Sicily and 290 kilometres north of Libya. Cominohas a permanent population of just 3, and with Gozo having apopulation of 31,143 persons, this leaves Malta as the major is-land of this archipelago state with 384,912 residents (Census 2011data) (National Statistics Office, 2014a).

Common examples of assistive technologies generallyused by older persons include telecare, smart toilets,wearable technologies, and smart homes. Telecare, asdiscussed in more detail in the subsequent section, in-cludes a telephone alarm and pendant at its most ba-sic, and wireless devices that measure home safety fea-tures and personal medical history at the most complex.Smart toilets are sites that conduct telehealth consulta-tion, with some even equipped with devices for bloodpressure and pathology tests on wastes. Wearable tech-nology consists of implanted devices such as heart pace-makers and artificial joints, and sensors embedded ina vest that measure cardio-respiratory and motion sig-nals. A smart home is a residence equipped with tech-nology that facilitates monitoring of residents aiming toimprove quality of life and physical independence otherthan to be a monitor of one’s health status. Herein, thetechnology becomes an element of the residential infra-structure and its performance does not rely on residentsreceiving training to operate it (Frisardi & Imbimbo,2011).

Assistive technology promises to taper the breakbetween individuals’ capacity and their environment,and hence, make it easier for people to remain in theirprevailing housing. Studies of impacts and outcomesof assistive technologies on the quality of life of olderpersons outline positive benefits in terms of quality oflife, safety and ease of performing everyday tasks (Bar-low, Singh, Bayer & Curry, 2007). In one particularstudy, users particularly valued the independence givento them by automatic showers, stair-lifts and downstairstoilets - as well as preventive benefits, which are likely toresult in reduced demands on health and social care ser-vices and reduced admissions to hospital and other com-munal care settings (Gramstad, Storli & Hamran, 2012).Assistive technologies also benefit persons with demen-tia (even those experiencing advanced stages of the dis-ease) - functioning to prevent, or at least delay, admis-sion to residential care for many people, whilst alsoproviding many carers with the reassurance, supportand peace-of-mind to continue in this capacity (Agree& Freedman, 2011). Carers appreciate being able toshare the caring duties, especially at night if the personfor whom they are caring is prone to wandering whenthe carer would normally be asleep. However, the degreeto which assistive technologies can succeed in reachingsuch a goal depends on older persons’ willingness to userelated equipment, something that in turn hinges on theneeds that people perceive, among which safety and theperceived usefulness of assistive technologies may be themost important.

10.7423/XJENZA.2015.1.01 www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16. 8

3 Telecare‘Telecare’ describes any service that brings health andsocial care directly to a user, generally in their ownhomes, and is supported by information and commu-nication technology. It is considered to be an innov-ative system for the delivery of health and care ser-vices that cuts across traditional domains of housing.An oft-cited definition states that ‘telecare is the re-mote or enhanced delivery of health and social servicesto people in their own home by means of telecommunica-tions and computerised systems” (Scottish Government,2009). Hence, “equipment and detectors that providecontinuous, automatic and remote monitoring of careneeds, emergencies and lifestyle changes, using informa-tion and communication technology. . . to trigger humanresponses, or shut down equipment to prevent hazards”(Scottish Government, 2009). In most cases, data is col-lected through sensors, fed into a smart residential huband sent electronically to a customer call centre.

In its course of development, there have been four ma-jor generations of Telecare systems (Brownsell, 2003).The first generation consisted of community alarms;simple technical systems that have no embedded intel-ligence, and are entirely reliant on the user activatinga call for help. The second generation has all the fea-tures of the first generation but also includes some levelof intelligence either locally in the home or dispersedthroughout the system (table 1). For example, sensorsmight be positioned both on the user and in the hometo detect alert situations and autonomously initiate acall for assistance if required. Hence, second generationtelecare systems are proactive.

Of course, it is unlikely that any user would need allthe components, so a range is available so that the mostappropriate can be chosen or prescribed. The home be-comes a safer environment in terms of appliances usedand detection of emergency situations, such as falls, fireor the presence of intruders. As needs change overtime, such as following early discharge from hospital,new components can be introduced and subscribed to.Overall, the second generation provides greater supportand monitoring in a way that addresses the particularneeds of service users.

The third generation adds further support capabilit-ies such as life-style monitoring, as well as contributingto an improvement in the user’s quality of life by sup-porting other tele-services such as banking, shopping,interactive exercise, medical diagnosis, and interactionwith other older people through teleconferencing andvideo conferencing (table 2). Tasks that were previ-ously carried out by staff but did not require contactwith users are taken over by the system. This there-fore enables staff to spend more time with users andto provide a more caring environment. Improvements

in medical monitoring will enable further parameters tobe measured, while for some users continuous 24-hourmedical monitoring will even be possible.

Finally, an emergent fourth generation involves theuse of Internet to deliver a vast range of telecare servicesthat overlap and integrate with other assistive technolo-gies (table 3). Whilst the emphasis of the previous threegenerations was to monitor users and detect emergencysituations, the fourth generation seeks to improve thesystem for both the user and provider. For instance,when service users are equipped with implanted sensorsunder the skin which communicate to an intelligent hub,fourth generation equipment provides health care pro-fessionals with ongoing monitoring and automatic as-sessment. Users can be formally assessed after the sys-tem has indicated that intervention would be beneficial.This allows resources to be effectively targeted, whilstalso allowing for a preventive system both in terms ofthe users’ health and their daily living.

Academia is only now beginning to attend to theviews and aspirations of prospective recipients of tele-care services, as well as conducting critical evaluations oftelecare packages. As one may expect, there is immenseenthusiasm and excitement accompanying accounts oftelecare’s potential usefulness, as a means of supply-ing care services in a cost effective manner, given theresource inadequacies of formal support services. Theconsensus in contemporary research is that telecare sys-tems have the potential to increase independence forolder people, support the delivery of more efficient ser-vices, and control expenditure on care which is pro-jected to increase massively in coming decades. TheUnited Kingdom government, for instance, has investedsignificantly in telecare development and delivery, andhas asserted that telecare could support older people toremain at home, reduce hospital admissions, improvequality of life, address workforce issues, cope with in-creasing demand, and save money (Bowes & McColgan,2013). In many respects, telecare was found to sup-port older people to realise their own goals, facilitatingtheir control over how they spent their time, and theirown determined efforts to retain independence. In termsof social participation and active citizenship, telecareprovided support to remain in one’s own home, henceproviding opportunities to enjoy social interactions ofpeople’s own choosing.

However, research has also questioned the capacityof telecare systems – thus, tempering the optimism ofpolicy makers. For example, Percival and Hanson (2006)- drawing on a series of focus groups involving serviceproviders, older people and family caregivers - identi-fied concerns about the possibility for telecare to reducehuman contact and thus increase isolation. Magnussonand Hanson (2003) identified complex ethical issues re-

10.7423/XJENZA.2015.1.01 www.xjenza.org

9 Formosa, M. (2015).Xjenza Online, 3:6–16.

Table 1: Key features of second generation telecare systems. Source: Brownsell (2003).

Main features Description

Fall detection Detect a fall and automatically instigate a call for assistanceFire detection Detect the presence of fire and inform user to contact centreGas detection Monitor levels of gas and keep the user informed as necessaryWater detection Suspend water supply if bath, shower or sink overflowIncontinence monitoring To indicate when incontinence pads need replacingSecurity Record arriving carers and disallow unauthorised accessDrug dispenser Dispense drugs when required and remind userMedical monitoring Measure medical characteristics and facilitate call for assistance

Table 2: Key features of third generation telecare systems. Source: Brownsell (2003).

Main features Description

Security Highly developed burglar alarms / automatic recognition of clientWeight detection Measures the client’s weight and outputs monthly chartsDrug dispenser Enables prescription reminders and remote analysis of medicationMedical band Provides 24-hour continuous medical monitoringDistance support Communication with medical centre and allows call for assistanceUser control Provides verbal communication with the home-based systemHealthy ageing Provides remote physiotherapy and exercise sessionsPharmacist Enables paperless prescription / delivery of prescribed medicine

Table 3: Key features of fourth generation telecare systems. Source: Brownsell (2003).

Main features Description

Water detection Monitoring the washing habits of the userGeneral assistance Mechanical assistance with cleaning / retrieving items from floorImplants Implanting sensors measuring vital signs 24-hours a day

garding telecare use, including its impact on complexfamily relationships, such as care partnerships in whichcare is coproduced. Similarly, Fisk (2003) highlightedhow some aspects of telecare are more focused on sur-veillance and monitoring than on promoting independ-ence and choice, raising important ethical issues. Otherresearch found that telecare service did not mitigateagainst ageist experiences, and service users still facedissues of ageism in many of the communities in whichthey lived (May, Mort, Williams, Mair & Gask, 2003).In other words, telecare provided older people with someprotection from acts of age discrimination but did notaddress them. As Dant (2006) suggested, the interac-tions of technical innovations with human relations mustbe a central concern for future evaluations of telecare.The use of the citizenship lens in evaluating telecare isimportant for focusing attention on the wishes of olderpeople themselves and draws attention away from theservice-led assessment of effectiveness.

In a study that explored the extent to which olderpeople, carers, and professionals consider telecare to be a

valuable/potentially valuable service, researchers foundthat service users draw attention to positive aspects oftelecare (Bowes & McColgan, 2013). However, data alsouncovered concerns relating to privacy and surveillance,and issues that highlight the interplay between socialcitizenship and individual freedom, that merit greaterattentiveness (Bowes & McColgan, 2013). Althoughmost carers made positive statements about the depth ofinformation that lifestyle monitoring and devices such aswandering alerts can generate, and the likely increasedknowledge they would subsequently have about a per-son’s risk levels at home, there was a general view thatthe data generated in respect of each individual has tobe subject to strict guidelines of confidentiality. In par-ticular, participants raised the spectre of commercialcompanies gaining lifestyle data and using this to directmarketing strategies to sell aids or adaptations. The im-portance of potential telecare service users maintainingand strengthening personal contact was also highlighted,usually by way of proclaiming that telecare should not‘take the place of face to face contact’, thereby reducing

10.7423/XJENZA.2015.1.01 www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16. 10

even more a person’s connection with a social world. Inthis respect, researchers urge policy makers to promotea holistic provision for Telecare services, one that allowsservice users to choose their preferred service and forthat service to be tailored to meet their changing needs.

4 The empirical universe

The second half of the 20th Century witnessed uniquedemographic changes. Declining fertility rates and mor-tality levels, and major improvements in life expectan-cies at birth, had far-reaching effects on global popu-lation trends, to the extent that the present epoch hasbeen referred to as the ‘age of ageing’ (Magnus, 2008).Malta is no exception to such trends. Figures based onthe 2011 Census indicate that, at end of 2013, 24.6 percent of the total Maltese population, or 105,068 persons,were aged 60-plus (National Statistics Office, 2014b).The number of persons aged 65-plus reached 76,024 oralmost 18 per cent of the population. The sex ratios forcohorts aged 65-plus and 80-plus numbered 79 and 55 re-spectively. Amongst cohorts aged 90-plus, the numberof older women was double that of men. Confirmingits belief that the institutionalisation of older personsin residential and nursing homes should only be a lastresort, the Government coordinates a number of com-munity services targeting older persons to enable themto live independently for as long as possible. These in-clude a card entitling its holder to obtain rebates andconcessions, handyman services, meals on wheels, homehelp services, incontinence services that supply heavilysubsidized diapers, telephone rebates to low income eld-ers, day centres, respite centres, and night shelters - inaddition to residential and nursing care homes (Parlia-mentary Secretariat for Rights of Persons with Disabil-ity and Active Ageing, 2013).

The Telecare Plus service owes its origins to the‘Telecare’ service which was established in 1991 as anemergency telephone service targeting older persons andothers with special needs. At the end of 2012, thetotal number of installations stood at 9,049 - with tele-operators handling some 100,000 calls every six months(Times of Malta, 2014). The aim of the Telecare servicewas to help older persons to continue living in their ownhomes for as long as possible, providing a 24- hour peaceof mind not only to older persons in their own homes,but also reassurance to family members and carers. Theclient was equipped with a pendant with an alarm sys-tem connected to the telecare centre. In case of an emer-gency the client would press the button that is found onthe telecare set, or the button on the pendant, which wasto be worn indoors, and kept within reach. Telephoneoperators would make contact with family members - ordoctors, health centres, and ambulances - in accordancewith the client’s needs. Research conducted by Cuta-

jar (2009) reported that subscribers to Telecare appliedfor this service because they were afraid that somethingwould happen to them while they were alone at home,believing that if there is an emergency the Telecare ser-vice would bring a quicker response than if they wereto use the normal telephone service. Clients were gen-erally positive about the service, reporting that it wasvery supportive and efficient, and it put their mind atrest. As regards whether the Telecare service should beimproved, clients suggested that the service would bebeneficial if instead of working with electricity it wouldwork on batteries, so that it would not stop working ifelectricity is cut off, whilst others suggested that theywould be better served if they could speak through thependant instead of having to reach the equipment in anemergency. Some also proposed that it would be muchmore beneficial if the service could be used outside thehouse within a certain distance.

The Telecare Plus service was launched in November2013, offering valuable add-ons to the previous Telecaresystem such as pill dispensers, and flood and gas detect-ors. The pendant can be upgraded to a ‘smart pendant’which also acts as a fall detector, intruder alarm, andmedication reminder. It is currently planned that thissmart platform will be upgraded in the future to buildup medical profiles of the person, who will be able tosubmit their blood pressure, diabetes or even peak flow,which will then trigger alerts to medical staff when theyare outside the upper or lower limit for that particu-lar patient. Eligible persons who can apply for such aservice include older couples/persons living alone, agedsixty years and over; persons with disability and thosewith special needs; and persons of any age who aretroubled by chronic systemic illnesses and who are livingalone and who are not gainfully occupied. The TelecareService is subsidised by state funds. No administrativefees are incurred in applying for this service. However,if the applicant is not (i) 60 years or over, (ii) in posses-sion of the pink form, (iii) living totally alone or withtwo or more elderly persons, he/she must pay the Tele-care Plus rental fee, which is e 4 a month. As at May2014, the number of all Telecare clients numbered 8,952:56 clients (Telecare and Carelink), 53 clients (Carelink),806 (Telecare Plus), and 8,037 (Telecare) (Departmentfor the Elderly and Community Care, unpublished stat-istics). Whilst 29.7 per cent of clients were male, asmuch as 69.7 per cent were female - 0.6 per cent of cli-ents were organisations. As regards living arrangements,17.1 per cent of clients were couples living alone, 2.3 percent were couples living with others, 69.5 per cent weresingle persons living alone, and 11.1 per cent were singlepersons living with others.

10.7423/XJENZA.2015.1.01 www.xjenza.org

11 Formosa, M. (2015).Xjenza Online, 3:6–16.

5 Research designThe study reported herein aimed to shed light on theimpact of Telecare Plus subscription on Maltese resid-ents aged 60-plus, seeking to uncover the extent that thisassistive technology contributes to their quality of life,and hence, its role in enabling them to ‘age-in-place’.The term ‘quality of life’ is an umbrella conceptualisa-tion that refers to wellbeing across various domains. Inprinciple, quality of life includes both subjective and ob-jective components, is based on individual needs, and iscomposed of multidimensional constructs influenced byrelationship and environmental factors (Agree & Freed-man, 2011). An oft-cited definition of the term ‘qualityof life’ is that promoted by the World Health Organiza-tion (1997), as “individuals’ perceptions of their positionin life in the context of the culture and value systemsin which they live and in relation to their goals, expect-ations, standards and concerns”. The objectives of thecase-study included the examination of the extent thatthe Telecare Plus service functions to improving - or atleast maintaining - service users’ emotional wellbeing,interpersonal relations, material wellbeing, personal de-velopment, physical wellbeing, self-determination, socialinclusion, and human rights.

The case study adopted a qualitative methodologywhich, in essence, “is concerned with how ordinarypeople manage their practical affairs in everyday life,or how they get things done”, constituting a “system-atic analysis of socially meaningful action through thedirect detailed observation of people in natural settingsin order to arrive at understandings and interpretationsof how people create and maintain their social worlds”(Neuman, 2002). Data collection was conducted bymeans of semi-structured interviews, and also via re-searcher’s observation of group activities and researchnotes, with the purpose of achieving triangulation. Insemi-structured interviews, the interviewer is normallyrequired to ask specific open-ended questions but is freeto probe beyond them if necessary with the interviewdeveloping as a joint product of what the intervieweesand interviewers talk about with each other. There-fore, semi-structured interviews contain the advantagesof both standardised and non-standardised interviewssuch as flexibility, control of the interview situation,and collection of supplementary information (Neuman,2002).

Informants for the semi- structured interviewswere selected through the process of non-probabilitysampling, which albeit making no claim to repres-entativeness, is a common strategy in exploratory re-search. The case study opted for a purposive typeof non-probability sampling where “researchers purpos-ively choose subjects who, in their opinion, are thoughtto be relevant to the research topic” (Sarantakos, 1993).

In this case, “the judgement of the investigator is moreimportant than obtaining a probability sample” (ibid.).Throughout the data collection process, efforts weremade in order to elicit information about service users’views on the impact that Telecare Plus is having ontheir quality of life. In between ice-breaker and closurequestions, interview enquiries focused on that interfacebetween the support awarded by Telecare Plus on onehand and quality of life on the other hand. Inspiredand adapted from research by Moe and Molka-Danielsen(2012) research, interview questions included:

• Do you trust that you can call for help and be heardwhen needed? Does use of Telecare Plus in yourhome help you enjoy life? Does use of Telecare Plusin your home make life more meaningful? (emo-tional wellbeing)

• Do you have daily individual communicationthrough Telecare Plus with your friends, family,support staff, and personal relations, in general?(interpersonal relations)

• How satisfied are you with your access to health ser-vices through Telecare Plus? How available is Tele-care Plus for access to your day-to-day informationtechnology needs? Does Telecare Plus improve yoursense of personal dignity? (personal development)

• Does Telecare Plus facilitate the lives of your in-formal carers? (physical wellbeing)

• Does Telecare Plus help you in your personal rela-tions? (social wellbeing)

Interviews were conducted with 26 older adults, allaged 60 years and above, with interviews ensuring un-failing adherence to the ethical principles (autonomy,beneficence, non-maleficence, justice) and rules (vera-city, privacy, confidentiality, fidelity) surrounding eth-ical conduct. The final stage consisted in the rigorousanalysis of data which resulted in the major themespresented in subsequent sections of this report. Datawas analysed through ‘logical analysis’ which first loc-ated premises within data that symbolise one group,and then explored connections between groups (Glaser& Strauss, 1967).

6 Results and Discussion

Of the 26 older people who took part in the research, 16lived alone and 10 with someone else, most with theirspouse. Whilst 8 were male, the remaining 18 were fe-male. All informants subscribed to the first generationservices of Telecare Plus. The reasons whereby researchparticipants chose to subscribe to Telecare were three-fold - namely, a modification in their living situation,such as becoming widowed and starting to live alone;changes in their health (often involving hospitalisation)in response to which social workers suggested a Telecaresubscription; and needing to put their ‘mind at rest’ if

10.7423/XJENZA.2015.1.01 www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16. 12

they were to experience an emergency. In all three trig-ger situations, social relationships were central to thedecision to subscribe to Telecare Plus. Family mem-bers were generally actively involved in organising theacquisition of this assistive technology service, usuallywith the aim of levelling their informal care responsib-ilities to a more manageable level. As one informantunderlined,

I am no longer young. . . On my return homefrom hospitalisation I was not as strong asbefore. I was afraid that I would fall again.I needed some reassurance that I could callfor help if something happened. . . My childrensuggested Telecare. They dealt with the ap-plication process. They were resolute that itwas either Telecare or applying for entry intoa care home! I do not blame them. They can-not stay with me all day long. They work andhave small children. . . All in all it was a gooddecision to apply for this service. It reassureseveryone. Help is just one push away, no needto fumble with telephone numbers. My eye-sight suffers these days.

Many research participants valued their regular par-ticipation in the community, especially attending reli-gious functions at the nearest parish church, and visit-ing day centres for the elderly. It was positive to notethat most enjoyed strong and supportive networks withneighbours in the community and family relatives. Thefact that Malta is a micro-state where no locality isfurther away than a 45-minute drive (excluding rushhours) certainly enables children to provide informalcare to their parents. Informants who claimed limitedability to leave their residences, particularly as a res-ult of mobility troubles, reported that almost all theirsocial interactions took place within their home, sup-plemented by visits to children’s and other family rel-atives’ residences on special occasions such as birthdaysand national holidays. As expected, the phone playeda central role in keeping contact with significant others,most claiming that they received a daily call from theirchildren. Only two subscribers owned a computer andinternet connection at home, with the rest of inform-ants being digitally illiterate. As a consequence, manyinterviewees professed to experience social isolation orloneliness, some suffering considerably from one or bothof these. Indeed, during the research interviews inform-ants frequently mentioned the relatively recent deathsand loss of their loved ones and friends, and the increas-ing number of acquaintances who were no longer ableto visit them due to frail health. In such circumstances,Telecare Plus filled an important void in older persons’lives. In an informant’s own words,

I am more alone at home than before. Manyof my friends cannot visit me as they cannotleave their homes or take public transport. Al-though I can leave my home, I cannot walklong distances. . . My children also visit me lessthan they used to do. . . The Telecare helps meto overcome my worries and anxiety. The timeswhen your family lived just round the cornerand were available all throughout the days arelong gone! The Telecare is good though, justone push of the button and you can speak tosomeone, I used it once when my oven stoppedworking, my son was here in less than an hour.

Research participants claimed that initially they per-ceived the Telecare Plus equipment as a cumbersomeaddition to their houses, especially when it was posi-tioned in the living room. They experienced stress andsubjective displeasure when facing the equipment as itreminded them of their frailty and vulnerability. How-ever, all informants stated that they eventually got usedto it, and even welcomed it in their homes once an emer-gency arose and they used the services of Telecare Pluswith satisfactory results.

6.1 Emotional wellbeing

Informants emphasised that Telecare Plus enables themto engage in a more active lifestyle in their communit-ies, supports them in carrying out everyday tasks, andeffectively manages risks to their health and home envir-onment. Even though some never had to use the facil-ities available through Telecare Plus, they felt confidentenough to stay living in the community, knowing that ifthey ran into some kind of difficulty then it was relat-ively easy and uncomplicated to ask for help. Presentedwith the queries ‘do you trust that you can call for helpand be heard when needed?’ and ‘does the use of Tele-care Plus in your home help you enjoy life and make lifemore meaningful?’, informants reacted as follows:

The Telecare makes it possible for people likeme to stay living in the community. Some-times I cannot sleep at night, I get afraid thatI am going to suffer another heart attack orthat someone will break into my house. Tele-care reassures me! I would have had to applyfor entry into a care home if it were not forTelecare.

Telecare enables me to enjoy my home, and dothe things I want to do, cook, phone my chil-dren, and receive visitors. I spend hours look-ing outside the window watching children goingto school, the cars go by, and the religious pro-cessions on Good Friday, Easter Sunday andduring the annual village feast. I know that

10.7423/XJENZA.2015.1.01 www.xjenza.org

13 Formosa, M. (2015).Xjenza Online, 3:6–16.

with Telecare help is just a call away. It helpsme to forget my physical problems and enjoythe day.

The contribution of Telecare Plus towards clients’emotional well-being was especially evident where ser-vice users found it difficult to use their telephone or anordinary mobile phone, either due to physical limita-tions such as arthritis and even early stage dementia.The equipment provided service users with a lifeline tocall for help in an emergency.

6.2 Interpersonal relations

Informants claimed that the Telecare Plus service func-tions as an interface that connects them on one hand,and family relatives and Telecare Plus operators on theother hand. In response to the questions ‘do you havedaily individual communication through Telecare Pluswith your friends, family, and support staff?’ and ‘doesTelecare Plus help you in your personal relations?’, in-formants underlined how the service provides them witha supportive network that links adequate social welfareand social aid in the community:

Being alone is the worst obstacle that olderpersons have to face in their daily lives. It isnot that you will not have anyone to speak tobut that if something happens to you, you arealways afraid that you will not be able to reachfor help. A few days ago I heard on the newsabout that old lady who was dead for two daysbefore her family relatives went to visit. I feelthat the key advantage of Telecare is that if Ifeel some physical ailment, especially at nightwhen I am in bed, I can use the pendant to callfor help. I do not have to get out of bed!

Telecare puts you in a system. You becomepart of the system. I did not like it at first. Irefused to admit that I am old and that I nowneed help. I think of myself more like a freebird if you know what I mean. . . I know that ifit were not for Telecare I would not be able tostay at home, and do all the things that I do.I would not be able to live with my wife. Howwould I be able to bear that?. . . Entering a carehome restricts you from living your life andmeeting the people you want to meet, peopleyou have been frequenting all your life.

During interviews, research participants oftenstressed the loss of strong interpersonal relations due tothe loss of friends, neighbours, and family members whohad died, or who were no longer able to visit because ofpoor health or disability. For frail users, their inabilityto leave their house meant that social interaction had

become limited to visitors coming to their homes. Asa result, their main challenges included loneliness andsocial isolation. Although the Telecare Plus enabledservice users to stay and live longer in the community,it was not contributing much to sustaining satisfactorylevels of interpersonal relations.

6.3 Personal development

The fact of getting used to the Telecare Plus equipment,experiences of using it in an emergency and dealing withalarms and alerts, was part of the lived experience of ser-vice users. It was positive to note that contrary to ste-reotypes that ‘older persons’ and ‘information techno-logy’ cannot live side-by-side, research data found oth-erwise - namely, that technology could really become‘domesticated’ and ‘embedded’ into the daily practicesof later life. Many described how despite initially notbeing very positive about subscribing to Telecare Plus,where such a subscription was to an extent forced uponthem by their family relatives (especially children), withtime they gained much confidence and were now evenfeeling comfortable with this form of assistive techno-logy. From responses to the questions ‘how available isTelecare Plus for access to your day-to-day informationtechnology needs?’ and ‘does Telecare Plus give youmore opportunity for leisure activity?’ it was evidentthat the service strengthened users’ self-esteem and theexecution of personal development plans:

Services such as Telecare allow older personssuch as myself to live in dignity. I never mar-ried, and as a result, do not have children. Ihave very minimal support, only that providedby neighbours and friends, most of whom areolder persons just as myself. Telecare is theonly alternative I have if I encounter an emer-gency. I do not have anybody to call in themiddle of the night. When I felt dizzy somemonths ago, I used Telecare and everythingwas sorted out for me by dawn. . . AlthoughI am old, I still want to do things in life, Istill want to live my life. Nothing much, mindyou, but for me even little things are import-ant. Telecare provides me with the support Ineed to continue living in my community andcontinue living my life.

It was also evident that when service users had friendswho also subscribed to Telecare Plus, the former do notfeel ‘different’ from significant others in their social net-work.

6.4 Social wellbeing

Telecare can lead to a variety of outcomes and con-sequences for family carers of older people. Whilst there

10.7423/XJENZA.2015.1.01 www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16. 14

is no doubt that caring for someone can be physically ex-hausting, to the extent that the physical health of carerstends to suffer greatly, constant worry also takes its tollon their mental health. Over half of the respondentsperceived that their subscription to Telecare Plus had apositive effect on their family carers. Reasons forwardedwere various and ranged from reduced stress, increasedself-assurance about the safety and welfare of the per-son cared for, having more prospects for respite fromcaring, positive developments in the relationship withperson cared for and family carer, to the capacity toremain in paid employment. In the informants’ words,

I want to continue living in my home as muchas possible. I want to be considered lucky ifI die here. But I do not want to be a bur-den on my children. My son and daughterhave demanding jobs and both have children.At one point they were taking it in turns tosleep with me. After some months, we realisedthat this was no longer possible. They were nolonger able to keep up with their lives. Tele-care made it possible, as well as safe, for meto sleep on my own. Although they still visitme every day, now they are able to get on withtheir lives with minimal disruption as possible.They can sleep soundly at home knowing thatif something happens to me, I can contact themthrough Telecare. Everybody feels better thisway.

My children are very happy that I acceptedthe Telecare service and that I know how itworks. They love me but they cannot live withme. And I do not want to live with them. Itried to live with them but the experiment onlylasted a week. We were all going mad! We livedifferent lives and we have different lifestyles.I am too old to change mine and they are tooyoung to accommodate my preferences. WithTelecare everybody is a winner, my childrensleep in their homes, I continue living at home,everybody’s mind is at rest that help is just apush of a button away. It was a good decision.Telecare is easy to operate.

Telecare brings to family carers a peace of mind anda strong bearing on their social inclusion - leaving themmore able to leave the house for any length of time to so-cialise, spend time with other family members or work- and most crucially, to dedicate some time to them-selves. Assistive technologies also function to preventcrises by preventing health complications, which mightresult in a loss of independence and the need for higherlevels of care provided by family members. Telecareprovides family carers the ability to stay in or return to

work, something that is not only a significant elementof improving carers’ future career prospects and socialinclusion but also has a direct impact on the likelihoodof family carers falling into debt and financial hardship.

6.5 Barriers

The research study highlighted a range of challenges to-wards an increased adoption of the Telecare Plus ser-vice by older people. From the tele-interviews, it wasclear that constrained user-friendliness may restrict theaptitude to control technological equipment, just as cog-nitive limitations can hinder the understanding of pro-cedures and navigation. Products are not always age-friendly - for example, directions may be in print, or inno print at all - so that instead of being empowering, as-sistive technologies may become the ‘last straw’ for anolder person. This may result in older people having alimited knowledge of possible products, and hence, theirfunctions and applications. As one informant explained,

I resisted the introduction of Telecare be-cause I took long to understand its functions.You hear Telecare this and Telecare that butnobody tells you exactly what it does. Nobodyshows it to you. I never saw the Telecare equip-ment until the day it entered my home. Noneof my friends or relatives had it. So it was abit too much for me to understand what it isand what it does. They should have demon-strations and adverts in television.

Lack of training on behalf of health professionals onthe use of technology, or the capacity to act on inform-ation, are also potential impediments. This may resultin some professionals resisting the further developmentand expansion of assistive technologies, which points tothe strong need for organisations involved in health andsocial care partnerships to develop approaches for pri-oritising client access to the types of assistive technolo-gies that they deliver. From the data it emerged thatother potential barriers include the perception on be-half of subscribers that they will have difficulty enga-ging with the technological requirements of the TelecarePlus equipment:

I resisted the introduction of Telecare becauseI was afraid that I would not be able to un-derstand how it works. I was always afraidof technology. I am not well-learned, I barelywent to school, I do not know how to drive, Ibarely can navigate the television set. . . I pre-ferred to complicate my life rather than get-ting some technological device which I wouldnot know how to operate. . . This was not wiseof me, but I am like that, I do not have much

10.7423/XJENZA.2015.1.01 www.xjenza.org

15 Formosa, M. (2015).Xjenza Online, 3:6–16.

confidence in myself. When you are illiterateyou prefer to keep a step back.

At the same time, most research participants indic-ated that they initially associated the use of TelecarePlus with a high degree of dependency and ill health.Wishing to distance themselves from negative connota-tions of old age, sickness and dependence, it followedthat they were not favourable to subscribe to the ser-vice. It is also possible that potential users believe thattheir quality of life will take a turn for the worse ifthey subscribe to an assistive technology in terms ofpotential threats to existing independence and servicearrangements. However, what is clear is that a lack ofinformation and discussions about prospects regardingthe technological interventions also leads to decisions todecline telecare services.

7 Conclusion

This study sought to conduct a systematic review ofthe home telecare service for vulnerable older people inMalta. Its goal was to examine the potential effects oftelecare in terms of benefits to individuals and carers.Results indicated that the key reasons whereby researchparticipants chose to subscribe to Telecare Plus werethree-fold: namely, a modification in their living situ-ation, such as becoming widowed and starting to livealone; changes in their health; and pressures on be-half of children so as for the latter to ‘put their mindat rest’. All things considered, the Telecare Plus ser-vice was contributing in exceptionally positive ways to-wards the quality of life of both subscribers and informalcarers, as exemplified by the statement of one of the in-formants: “With Telecare everybody is a winner, mychildren sleep in their homes, I continue living at home,everybody’s mind is at rest that help is just a push of abutton away” (male, 70 years old).

Assistive technologies such as Telecare Plus have thepotential to empower older persons to continue liv-ing at home if that is their preference, although it isnoteworthy that there will always be some compromisebetween meeting the desired sense of independence andan enduring element of risk in living at home. At theforefront of benefits for users, telecare systems can alle-viate the anxiety caused by age or long-term health con-ditions, whilst also improving an older person’s sense ofsecurity and self-confidence. Moreover, since most avail-able telecare systems have the possibility to be upgradedto incorporate second, third and fourth generation ser-vices, the level of telecare provision can be increased asnew health problems develop. Telecare systems also al-leviate some of the burdens and anxieties that affect in-formal carers, and hence, serve to improve their qualityof lives too. This is highly positive as it can encourage

family members to carry on caring for longer, which cancircumvent the older person moving into a care home forlonger periods. Although not as a result of this study, itis evident that third and fourth generation telecare tech-nologies assist in caring for older people suffering fromlong-term chronic conditions such as diabetes, asthmaand high blood pressure. Hence, this encourages olderpeople with these conditions to remain living at home,while also monitoring any further deterioration in theircondition. This means that telecare systems also havethe potential to be cost-effective by circumventing orpostponing a person’s move into a care home or hospital.Indeed, telecare can hasten an older person’s dischargefrom hospital by providing added support in their ownresidence or in another intermediate care setting, thusfreeing up hospital beds.

The research data elicited possible barriers to the takeup of telecare services in Malta. Indeed, although thereexists an overabundance of innovative forms of assistivetechnologies, the uptake is far from ideal. Most pre-occupying was the fact that none of the informants inthe convenience sample subscribed to second, third andfourth generation versions of Telecare services. In ad-dition to the potential difficulties listed in the resultsand discussion section, especially the constrained user-friendliness, the fact that the field of assistive techno-logies lacks efficient and clear business models consti-tutes a key limitation towards the take up of such ser-vices. The adoption rates of assistive technologies arevery low in most countries, with very few people liv-ing in smart homes, and with the available ones gener-ally existing for research and demonstration purposes.This occurs despite the fact that the need for such tech-nology is apparent through the predominance of issuesranging from wandering, leading to getting lost and self-harm, accidental falls, medication complications, contin-ence management, and the high fear of crime amongstolder persons. Indeed, one missing element constitutes abusiness process for assessment, specification of suitabletechnology, installation, and maintenance, and a serviceto monitor the signals. At the same time, existing busi-ness models are wrong to project assistive technologiesas the solution to all social and health care ailments, andit remains necessary that assistive technology is clearlyrecognized as only one component of a care package.For instance, technological systems for monitoring a per-son’s vital signs are to be linked to systems and protocolsfor providing assistance when required. Moreover, givenits holistic approach to health care, strategies for in-creasing population access to assistive technologies needto be developed by, and have the support of, all stake-holders responsible for the welfare of older and ageingpeople.

10.7423/XJENZA.2015.1.01 www.xjenza.org

Formosa, M. (2015).Xjenza Online, 3:6–16. 16

AcknowledgementI am indebted to Tessa Fiorini-Cohen for reading andreacting to an earlier draft of this research article.

References

Agree, E. M. & Freedman, V. A. (2011). A Quality-of-Life Scale for Assistive Technology: Results of aPilot Study of Aging and Technology. Phys. Ther.91 (12), 1780–1788.

Barlow, J., Singh, D., Bayer, S. & Curry, R. (2007). Asystematic review of the benefits of home telecarefor frail elderly people and those with long-termconditions. J. Telemed. Telecare, 13 (4), 172–179.

Bowes, A. & McColgan, G. (2013). Telecare for OlderPeople: Promoting Independence, Participation,and Identity. Res. Aging, 35 (1), 32–49.

Brownsell, B. (2003). Assistive Technology and Telecare:Forging Solutions for Independent Living. PolicyPress, Bristol.

Cutajar, M. D. (2009). Community-based Services forOlder Persons. (Doctoral dissertation, Universityof Malta).

Dant, T. (2006). Material civilization: Things and soci-ety. Br. J. Sociol. 57 (2), 289–308.

Fisk, M. J. (2003). Social Alarms to Telecare: OlderPeople’s Services in Transition (1st). Bristol:Policy Press.

Frisardi, V. & Imbimbo, B. P. (2011). Gerontechno-logy for demented patients: Smart homes for smartaging. J. Alzheimer’s Dis. 23 (1), 143–146.

Glaser, B. G. & Strauss, A. L. (1967). The Discovery ofGrounded Theory: Strategies for Qualitative Re-search. Chicago: Transaction Publishers.

Gramstad, A., Storli, S. L. & Hamran, T. (2012). “DoI need it? Do I really need it?” Elderly peoplesexperiences of unmet assistive technology deviceneeds. Disabil. Rehabil. Assist. Technol. 1–7.

Horner, B., Soar, J. & Krich, B. (2009). Assistive tech-nology: Opportunities and implications. In R. Nay& S. Garratt (Eds.), Nurs. older people issues in-nov. (3rd ed., Chap. 23). Sydney: Elsevier.

Magnus, G. (2008). The Age of Ageing: How Demo-graphics are Changing the Global Economy andour World. Hoboken, New Jersey: Wiley.

Magnusson, L. & Hanson, J. (2003). Ethical issuesarising from a research, technology and develop-ment project to support frail older people andtheir family carers at home. Health Soc. CareCommunity, 11 (5), 431–439.

Mann, W. C. (2005). Smart Technology for Aging, Dis-ability, and Independence: The State of the Sci-ence. Hoboken, New Jersey: Wiley.

May, C., Mort, M., Williams, T., Mair, F. & Gask, L.(2003). Health technology assessment in its localcontexts: Studies of telehealthcare. Soc. Sci. Med.57 (4), 697–710.

McCreadie, C. & Tinker, A. (2005). The acceptabilityof assistive technology to older people. Ageing Soc.25 (1), 91–110.

Moe, C. E. & Molka-Danielsen, J. (2012). IndependentLiving for the Elderly: Development of an Assess-ment Framework for Comparison of Assistive ICTInitiatives. Nor. Konf. Organ. Bruk av Informas-jonsteknolo, (January), 265.

National Statistics Office. (2014a). Census of Popula-tion and Housing 2011. Volume 1: Population. Na-tional Statistics Office. Malta.

National Statistics Office. (2014b). World PopulationDay 2014. National Statistics Office. Malta.

Neuman, L. W. (2002). Social Research Methods: Qual-itative and Quantitative Approaches (5th ed.). Bo-ston: Allyn & Bacon.

Parliamentary Secretariat for Rights of Persons withDisability and Active Ageing. (2013). NationalStrategic Policy for Active Ageing: Malta 2014-2020. Parliamentary Secretariat for Rights of Per-sons with Disability and Active Ageing. Malta.

Percival, J. & Hanson, J. (2006). Big brother or bravenew world? Telecare and its implications for olderpeople’s independence and social inclusion. Crit.Soc. Policy, 26 (4), 888–909.

Sarantakos, S. (1993). Social Research (1st). Australia:Macmillan Education.

Scottish Government. (2009). National Telecare Devel-opment Programme in Scotland: Glossary of termsand definitions.

Sixsmith, A. & Gutman, G. (2013). Technologies forActive Ageing (A. Sixsmith & G. Gutman, Eds.).New York: Springer.

Times of Malta. (2014). Telecare handling 100,000 callsevery six months. Retrieved January 30, 2015,from http : / / www . timesofmalta . com / articles /view/20140130/business-news/Telecare-handling-100-000-calls-every-six-months.504799

World Health Organization. (1997). Measuring Qualityof Life: The World Health Organization Qualityof Life Instruments. World Health Organization.Geneva.

10.7423/XJENZA.2015.1.01 www.xjenza.org

Baisch, U. (2015).Xjenza Online, 3:17–22.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.02

Review Article

Connecting Frontier Research with Industrial Development - Lanthan-ide and Actinide Chemistry in the European f-Element Network(EUFEN)

U. BaischDepartment of Chemistry, University of Malta, MSD 2080, Msida, Malta

Abstract. The chemistry of f-elements, even thoughoften not known to the general public, forms part ofmany industrial processes, manufacturing and medicalapplications, such as medical imaging (e.g. MRI scans),strong magnets, data storage media, superconducters,LEDs, catalysis, as well as energy, and metal extrac-tion. It plays a crucial role in the scientific and industriallandscape of the European Union (EU) in fields of en-ergy, security, training, sustainability, and society. Theuse of these elements widens the scope of synthetic pos-sibilities in chemistry, and materials with outstandingelectromagnetic properties have already been realised.The synthesis of lanthanide containing supramolecu-lar materials with exceptional materials properties hasalready been reported, e.g. the ability to bind andrelease gases, high-temperature superconductivity, andall-white light emitting diodes. The EUFEN (Europeanf-Element Network) COST action provides cooperat-ive mobility mechanisms for nationally funded f-elementchemists pursuing fundamental frontier research to initi-ate collaborations, training, networking, and dissemina-tion among each other. Novel developments and resultsin terms of f-element crystal engineering carried out atthe University of Malta are therefore part of EUFEN.

1 Introduction

EUFEN, the European f-Element Network is a COSTaction which provides cooperative mobility mechanismsfor f-element chemists from all over Europe for pursuingfundamental frontier research to initiate collaborations,training, networking, and dissemination with each other(COST, 2014).

Two goals for the action have been identified: (i) totackle unsolved problems in f-element chemistry and (ii)

to supply industry with trained researchers from univer-sities, restructuring the EU science-base, and thus resultin a research output that is greater than the sum of itsparts.

Although already more than 200 years old, the chem-istry of f-elements (a block of 28 elements at the bottomof the periodic table) can still be considered as one theof areas of chemistry that are least known by the gen-eral public. Hardly anybody outside the field of inor-ganic chemistry will have read or heard of “f-elements”as well as groups of elements called lanthanides, actin-ides or rare earths; a good example are the actinide ele-ments uranium or plutonium: even though one mightknow these elements very well in the context of radio-activity or nuclear energy, only a small group of peoplewould know that uranium and plutonium are part of thegroup of elements in the periodic table called f-elementsand that there is a surprisingly rich amount of syntheticchemistry of e.g. uranium and thorium complexes pub-lished every year (Bunzli, 2006); a search with the pro-gram SciFinder using the key words thorium or uraniumcomplex resulted in over 5000 publications since 2000.

The innocence of the existence of a whole group of28 elements is somewhat surprising as f-elements (inparticular lanthanides) are indispensable components inmany materials everybody uses day by day. f-Elementchemistry contributes to medical imaging (MRI contrastagents), magnetic (strong magnets), electronic (super-conductors), and photonic devices (lasers, phosphors,displays, LEDs), catalysis, energy, and metal extractionand is therefore strategically crucial to science, energy,security, training, sustainability, and society (Blake etal., 1999; Bunzli, 2006; Bunzli & Piguet, 2002; Edel-mann, 2009). This wide range of applications is theresult of the particular electromagnetic properties of

Correspondence to: U. Baisch (ulrich.baisch@um.edu.mt)

c© 2015 Xjenza Online

Baisch, U. (2015).Xjenza Online, 3:17–22. 18

f-elements. Often just a very small percentage of f-element (usually between 0.01 and 10 %) needs to bepresent in a material to change its optical, electrical ormagnetic properties completely.

The synthesis of lanthanide containing supramolecu-lar materials with exceptional materials properties havebeen discovered recently, e.g. the ability to bind and re-lease gases, superconductivity, and luminescence (Blakeet al., 1999).

In addition to the general lack of knowledge aboutthe existence of these elements in public, there are alsoa number of ways to name certain groups of f-elementsand other related elements in the periodic table. Ele-ments called lanthanides and actinides can either i) bethe row of elements from lanthanum (La) to ytterbium(Yb) and actinium (Ac) to nobelium (No) or ii) includealso the elements lutetium (Lu) and (Lr) lawrencium,respectively. However, according to the conventions setby the International Union of Pure and Applied Chem-istry (IUPAC), the latter (ii) would refer to these groupsas lanthanoids and actinoids rather than lanthanidesand actinides. Some sources also state that the groupof lanthanides consists of the group of elements fromcerium (Ce) to lutetium (Lu) (Bunzli, 2006). Rareearths is also an expression which is often used for thegroup of 4f and 5f-elements. This group consists notonly of 4f and 5f elements, but it includes also otherelements from Group 3 of the periodic system, namelyyttrium (Y) and scandium (Sc).

In view of their large ionic radius and their partic-ular electronic configuration, f-element based researchcan be extremely challenging from a synthetic point ofview. Scientific exchange and collaboration are thereforeessential in this field of research. EUFEN offered for thefirst time a platform for f-element researchers from allover Europe to join forces and tackle the pending ques-tions, such as: What are the principles/mechanisms be-hind the use of lanthanide complexes as catalysts? Howcan f-element compounds be treated in computationalchemistry and how can we predict the formation of newcomplex compounds (COST, 2014)?

Due to their special physical and chemical properties,rare earth elements were almost unused and less popularin the field of crystal engineering (Broker, Klingshirn& Rogers, 2002). Novel developments and results interms of f-element crystal engineering are therefore alsoan important part of EUFEN.

As EUFEN is the first and only collaborative researchnetwork in the field of f-element chemistry, a very largepart of the European community of f-element scientistsis involved. Over 120 researchers from 24 nations formpart of this COST action. Three main working groupswere set up (COST, 2014):

a) Synthesis and Structure: the aim is to undertake

the synthesis, structures, and chemical bonding in new f-element compounds, with the scope of discovering novelstructures and bonding and rationalising the observedphenomena.

b) Spectroscopy and Computation: the aim is to fo-cus on exploiting existing and new f-element compoundsfor novel spectroscopic and computational investigationswhich hitherto have no precedent.

c) Applications: the aim is to explore existing andnew chemical bonding and reactivity, or other phenom-ena, with a view to ultimately delivering research whichcan benefit EU science, industry, and society as oppor-tunities to exploit new findings become available.

Three different major published outcomes of collabor-ative EUFEN research will be described below in orderto showcase the COST action. Subsequently, our ownscientific contribution will be summarised.

2 Supramolecular Sensors

One of the most prominent properties of both 4f- and5f-elements is their large atomic or ionic size. As a con-sequence their metal ions can usually coordinate to ahigher amount of ligands (coordination numbers varybetween 7 and 12) than is usually the case for transitionmetals. The geometries around the f-element centre areless restricted and this makes 4f-element complexes inparticular a well-known component in supramolecularchemistry.

Three research participants of EUFEN (Universit-ies of Strasbourg and Bretagne Occidentale, France;University of Coruna, Spain) collaborated on a pro-ject which examined the effect of fluoride addition toaqueous solutions of luminescent lanthanide complexes(Ln = Eu, Tb, Yb).(Liu et al., 2014) Upon addition,the formation of a dimeric europium complex Eu-1 wasobserved. X-ray crystal structure determination and lu-minescence spectroscopy revealed not only the geometryof the complexes but also that i) fluoride anions wereconfined into the cavity that was formed by the two com-plex molecules and ii) luminescence intensity increasedsignificantly. Synergistic effects of the Eu-F-Eu bridgingmotif, π stacking interactions, and a four-componenthydrogen-bonding network which control the assemblyof the two complex molecules around the fluoride ion,are crucial for the above (see Figure 1).

The exact sensing of fluoride in aqueous solutions orsolvent mixtures is of great importance for public health(Liu et al., 2014). Fluoride in small quantities can havea positive impact on teeth and bones, whereas higherconcentrations of fluoride can result in serious healthproblems. According to the World Health Organisationthe fluoride content in drinking water should be lowerthan 1.5 ppm (World Health Organization, 2004). How-ever, so far there are just a few procedures known to

10.7423/XJENZA.2015.1.02 www.xjenza.org

19 Baisch, U. (2015).Xjenza Online, 3:17–22.

Figure 1: Crystal structure of the Eu-1 dimer viewed per-pendicular to (a and b) the main pseudo-C2 axis.(Liu et al.,2014)

quantitatively analyse the fluoride content in aqueoussolutions.

The authors’ new approach is to use the lanthanidecomplexes described above to sense the fluoride con-tent in aqueous solutions using fluorescence spectro-scopy (Liu et al., 2014).

3 Single Molecule Magnets

Single-molecule magnets (SMMs) are metal-organiccompounds that exhibit paramagnetic behaviour on amolecular level. They can be described as the smal-lest possible magnetic devices. Potential applications ofSMMs are quantum computing, high-density informa-tion storage and magnetic memory devices. Lanthanide-SMMs play a leading role in this field of research as thepresence of 4f-elements in a complex molecule alters themagnetic behaviour drastically due to the presence of4f-electrons and thus show a particularly high magneticmoment and single ion anisotropy compared to trans-ition metal SMMs (Christou, Gatteschi, Hendrickson &Sessoli, 2000). However, also systematic research in thefield of actinoid (5f) based SMMs emerged in the liter-ature, recently. The outcomes discussed in these papershave resulted from STSMs (Short-Term Scientific Mis-sion) of EUFEN (Liddle, Mills & Wooles, 2010, 2011;Meihaus & Long, 2015; Mougel et al., 2012, 2012).

Okuda (RWTH Aachen, Germany), Layfield (Univer-sity of Manchester, UK) and coworkers have collabor-ated within EUFEN and carried out an experimentaland ab initio computational study of an asymmetrical,hydride-bridged di-dysprosium single-molecule magnet(Venugopal et al., 2013).

Hydride ligands can be of particular interest forthe development of SMMs as they exhibit very strongligand-field effects. Subsequently, this could have a dif-ferent effect on the relaxation times in Ln-SMMs com-pared to metal-organic compounds with oxygen-donorligands (Luzon & Sessoli, 2012; Rinehart & Long, 2011;Sorace, Benelli & Gatteschi, 2011).

The reported compounds in this study werethe first hydride-ligated SMMs. The com-

Figure 2: Orientation of the magnetic moments in Ln-2 (dashed lines). The arrows show the antiferromagneticcoupling. Pink atoms = hydride atoms, blue = N atoms,grey = C atoms (Venugopal et al., 2013).

pounds consist of hydride bridged complexes[Ln(Me6trenCH2)(µ−H)3Ln(Me6tren)][BC6H3(CF3)24]2Ln-2,where Ln = Gd(III), Dy(III) and Me6tren =tris[2-(dimethylamino)ethyl]amine (see Figure 2).

4 Uranium(VI)–nitride triplebond

The chemistry of uranium is not only related to its roleas a radioactive element, but it is in fact an elementwell known for its extraordinary coordination chemistry(Bart & Meyer, 2008; Hayton, 2010; King & Liddle,2014; Lu, 2014; Van Horn & Huang, 2006). Uraniumcan exist in different oxidation states and can form com-plexes with high coordination numbers around the ac-tinide (5f) centre due to its large atomic and ionic size.

Whereas the bonding of ligands to 4f elements(lanthanides) is generally known to be mainly ionic,the nature of 5f-element (actinides) ligand bonds is stilldiscussed in recent literature. One of the most con-troversially analysed aspects of this is the existence ofuranium-ligand multiple bonds (Baker, 2012; Bart &Meyer, 2008; Hayton, 2010; King & Liddle, 2014).

During the past years uranium(V)-OR and –NR werereported (R = alkyl group) and thus, the synthesis ofa uranium-ligand triple bond was one of the hot topicsof EUFEN as it is of fundamental importance to thestudy of f-orbital participation in metal-ligand multiplebonding (Hayton, 2013).

It was therefore a big break-through when Liddle(University of Nottingham, UK), McInnes (Univer-sity of Manchester, UK) and coworkers reported thepreparation of a complex with a uranium(VI)–nitridetriple bond: the terminal uranium(V) nitride com-plex [UN(TrenTIPS)][Na(12-crown-4)2] (TrenTIPS =[N(CH2CH2NSiiPr3)3]3 - and iPr = isopropyl) U-3.Synthesis was achieved as summarized in Figure 3 (Kinget al., 2012, 2013).

The existence of a triple bond was proven by single-

10.7423/XJENZA.2015.1.02 www.xjenza.org

Baisch, U. (2015).Xjenza Online, 3:17–22. 20

Figure 3: Reaction scheme for the synthesis of multiple uranium-nitrogen bonds (U-3) (King et al., 2013).

crystal X-ray diffraction. The uranium-terminal nitridebond length of 1.825 A is significantly shorter than thedistances reported for uranium amides and uraniumamines, which average at around 2.33 A and 2.66 A (Lu,2014).

A thorough understanding of the nature of uraniumligand bonds could result in new developments fornuclear-waste clean-up, as well as new applications inthe field of catalysis or in the synthesis of polymericuranium nitride [UN]n for use as a ceramic nuclear fuel(Streit & Ingold, 2005).

5 Lanthanide Crystal Engineer-ing in Malta

Solubility is one of the biggest challenges industrialchemistry has to face day by day. Prominent examplesare paints, coatings, pigments, and last but not least,pharmaceutically active compounds. Almost all of thesecompounds are obtained as a solid at some stage duringthe manufacturing process. Solids are often preferredas the final product because of the obvious advantagesin transporting and storing. However, almost all of theabove-mentioned products are applied in the form ofsolutions or suspensions: e.g. drugs are swallowed as asolid and then will have to be dissolved in the stomachto enter the blood stream; paints have to be applied assolutions or suspensions.

Thus, almost all of the solids produced will have to bedissolved or suspended again when used. The solubilityof a solid depends mainly on the amount of attractionbetween the individual molecules, ions or atoms in thecrystal lattice, which again depends on their 3D arrange-ment. Crystal chemistry of molecular networks, wheremolecules are linked to each other by non-covalent inter-

actions (H-bonds, π-interactions) or coordinative bondsinvolving multidentate ligands, can be considered as oneof the most promising and appealing branches in mod-ern solid-state chemistry. The formation and recom-bination of these molecular building blocks permits aunique fine-tuning of the physical-chemical propertiesof the synthesized compound in the solid-state withoutaltering the actual molecule itself.

During the past decade a new field of science emergedfrom the above-mentioned necessity to control the ar-rangement of molecules and ions in crystalline solids:Crystal Engineering. It is “the understanding of the in-termolecular interactions in the context of crystal pack-ing and the utilization of such understanding in thedesign of new solids with desired physical and chem-ical properties” (Desiraju, 1989). This area of researchdoes not fall neatly within the classical fields of chem-istry (organic chemistry, inorganic chemistry, physicalchemistry, materials chemistry, biochemistry). It cutsacross these traditional vertical subdivisions. Analyt-ical knowledge (e.g. crystallisation techniques, X-raydiffraction, biological processes, and thermo-analyticalmethods) and the practical experience to synthesiseorganic, complex, and organometallic compounds areequally important.

In spite of its great potential, very few research groupshave applied a crystal engineering approach to molecu-lar metal-based compounds in general (Braga, Grepi-oni & Maini, 2010; Evans & Lin, 2002; Kanaizuka etal., 2008; Mınguez Espallargas et al., 2010; Zaworotko,2006). Their inclusion into cocrystal and polymorphscreening for the development of the treatment is stillto be exploited both from a theoretical and practicalpoint of view.

This is true in particular in the field of 4f-element

10.7423/XJENZA.2015.1.02 www.xjenza.org

21 Baisch, U. (2015).Xjenza Online, 3:17–22.

Figure 4: Crystal structure of Yb-4 with view along cellaxis b, dashed lines depict hydrogen bonding (Baisch &Braga, 2009).

chemistry. Lanthanide elements are chemically verysimilar to alkaline and alkaline earth elements and thuscan also influence the solubility of a compound signi-ficantly. When surrounded by ligands, these elementshave very low toxicity apart from being generally verystable towards oxidation.(Dyson & Sava, 2006; Mewis& Archibald, 2010) They could be utilised effectivelynot only in the formation of cocrystals or salts to al-ter the physical properties of drugs, but also to altercrystal growth/solubility of solids which are harmful tothe body (e.g. urate crystals). There is currently noprecedent literature for this approach.

Our group at the University of Malta functions asa collaborator in all aspects of Crystal Engineering forvarious 4f-element complexes (ionic and neutral) in theEUFEN COST action. Polymorph screening is carriedout in order to explore the existence of new crystal formswith potentially different solubility properties.

The synthesis of lanthanide containing supramolecu-lar compounds with exceptional materials properties hasalready been achieved as can be seen below (Figure 4)(Baisch & Braga, 2009). This compound has been syn-thesized by the reaction of melamine with ytterbium ox-alate Yb-4 in boiling water. Single-crystal X-ray ana-lysis of suitable crystals revealed a structure in whichmelamine forms layers via an extended hydrogen bondnetwork. The lanthanide complex molecules connectthese layers vertically by multiple hydrogen bonds.

References

Baisch, U. & Braga, D. (2009). [Yb(C2O4)4]5− - a ver-satile metal-organic building block for layered co-ordination polymers. Cryst. Eng. Comm. 11 (1),40–42.

Baker, R. J. (2012). The coordination and organometal-lic chemistry of UI3 and UN(SiMe3)23: Syn-thetic reagents par excellence. Coord. Chem. Rev.256 (23-24), 2843–2871.

Bart, S. C. & Meyer, K. (2008). Highlights in uraniumcoordination chemistry. Struct. Bond. 127, 119–176.

Blake, A. J., Champness, N. R., Hubberstey, P., Li,W. S., Withersby, M. A. & Schroder, M. (1999).Inorganic crystal engineering using self-assemblyof tailored building-blocks. Coord. Chem. Rev.183, 117–138.

Braga, D., Grepioni, F. & Maini, L. (2010). The growingworld of crystal forms. Chem. Commun. 46 (34),6232–6242.

Broker, G. A., Klingshirn, M. A. & Rogers, R. D. (2002).Green chemistry and lanthanide-based crystal en-gineering. J. Alloys Compd. 344 (1-2), 123–127.

Bunzli, J. C. G. (2006). Benefiting from the UniqueProperties of Lanthanide Ions. Acc. Chem. Res.39, 53–61.

Bunzli, J. C. G. & Piguet, C. (2002). Lanthanide-containing molecular and supramolecular poly-metallic functional assemblies. Chem. Rev. (Wash-ington, D. C.) 102, 1897–1928.

Christou, G., Gatteschi, D., Hendrickson, D. N. &Sessoli, R. (2000). Single-molecule magnets. MRSBull. 25 (11), 66–71.

COST. (2014). COST Action CM1006: EUFEN.Desiraju, G. R. (1989). Crystal engineering: the design

of organic solids. Amsterdam: Elsevier.Dyson, P. J. & Sava, G. (2006). Metal-based antitumour

drugs in the post genomic era. Dalt. Trans. 35 (16),1929–1933.

Edelmann, F. T. (2009). Lanthanides and actinides: An-nual survey of their organometallic chemistry cov-ering the year 2006. Coord. Chem. Rev. 253 (3-4),343–409.

Evans, O. R. & Lin, W. (2002). Crystal engineering ofNLO materials based on metal–organic coordina-tion networks. Acc Chem Res, 35 (7), 511–522.

Hayton, T. W. (2010). Metal-ligand multiple bondingin uranium: structure and reactivity. Dalt. Trans.39 (5), 1145–1158.

Hayton, T. W. (2013). Recent developments in actinide-ligand multiple bonding. Chem. Commun. 49 (29),2956–2973.

Kanaizuka, K., Haruki, R., Sakata, O., Yoshimoto, M.,Akita, Y. & Kitagawa, H. (2008). Constructionof Highly Oriented Crystalline Surface Coordin-ation Polymers Composed of Copper Dithioox-amide Complexes. J. Am. Chem. Soc. 130 (47),15778–15779.

King, D. M. & Liddle, S. T. (2014). Progress in molecu-lar uranium-nitride chemistry. Coord. Chem. Rev.266-267, 2–15.

10.7423/XJENZA.2015.1.02 www.xjenza.org

Baisch, U. (2015).Xjenza Online, 3:17–22. 22

King, D. M., Tuna, F., McInnes, E. J. L., McMaster,J., Lewis, W., Blake, A. J. & Liddle, S. T. (2012).Synthesis and Structure of a Terminal UraniumNitride Complex. Science (80). 337 (6095), 717–720.

King, D. M., Tuna, F., McInnes, E. J. L., Mc-Master, J., Lewis, W., Blake, A. J. & Liddle,S. T. (2013). Isolation and characterization ofa uranium(VI)–nitride triple bond. Nat. Chem.5 (6), 482–488.

Liddle, S. T., Mills, D. P. & Wooles, A. J. (2010).Bis(phosphorus-stabilised)methanide and meth-andiide derivatives of group 1-5 and f-elementmetals. Organomet. Chem. 36, 29–55.

Liddle, S. T., Mills, D. P. & Wooles, A. J. (2011). Earlymetal bis(phosphorus-stabilised)carbene chem-istry. Chem. Soc. Rev. 40 (5), 2164–2176.

Liu, T., Nonat, A., Beyler, M., Regueiro-Figueroa,M., Nchimi Nono, K., Jeannin, O., . . . Char-bonniere, L. J. (2014). Supramolecular Lumines-cent Lanthanide Dimers for Fluoride Sequesteringand Sensing. Angew. Chem. Int. Ed. Engl. 53 (28),7259–7263.

Lu, Y. (2014). Coordination chemistry in the ocean. Nat.Chem. 6 (3), 175–177.

Luzon, J. & Sessoli, R. (2012). Lanthanides in molecu-lar magnetism: so fascinating, so challenging. Dalt.Trans. 41 (44), 13556–13567.

Meihaus, K. R. & Long, J. R. (2015). Actinide-basedsingle-molecule magnets. Dalt. Trans. ahead ofprint, DOI: 10.1039/C4DT02391A.

Mewis, R. E. & Archibald, S. J. (2010). Biomedical ap-plications of macrocyclic ligand complexes. Coord.Chem. Rev. 254 (15-16), 1686–1712.

Mınguez Espallargas, G., van de Streek, J., Fernandes,P., Florence, A. J., Brunelli, M., Shankland, K.& Brammer, L. (2010). Mechanistic Insights intoa Gas–Solid Reaction in Molecular Crystals: TheRole of Hydrogen Bonding. Angew. Chem. Int. Ed.Engl. 49 (47), 8892–8896.

Mougel, V., Chatelain, L., Pecaut, J., Caciuffo, R.,Colineau, E., Griveau, J.-C. & Mazzanti, M.(2012). Uranium and manganese assembled in awheel-shaped nanoscale single-molecule magnetwith high spin-reversal barrier. Nat. Chem. 4 (12),1011–1017.

Rinehart, J. D. & Long, J. R. (2011). Exploiting single-ion anisotropy in the design of f-element single-molecule magnets. Chem. Sci. 2 (11), 2078–2085.

Sorace, L., Benelli, C. & Gatteschi, D. (2011). Lanthan-ides in molecular magnetism: old tools in a newfield. Chem. Soc. Rev. 40 (6), 3092–3104.

Streit, M. & Ingold, F. (2005). Nitrides as a nuclear fueloption. J. Eur. Ceram. Soc. 25 (12), 2687–2692.

Van Horn, J. D. & Huang, H. (2006). Uranium(VI) bio-coordination chemistry from biochemical, solutionand protein structural data. Coord. Chem. Rev.250 (7-8), 765–775.

Venugopal, A., Tuna, F., Spaniol, T. P., Ungur,L., Chibotaru, L. F., Okuda, J. & Layfield,R. A. (2013). A hydride-ligated dysprosium single-molecule magnet. Chem. Commun. 49 (9), 901–903.

World Health Organization. (2004). WHO: Flouride inDrinking Water.

Zaworotko, M. J. (2006). Molecules to Crystals, Crystalsto Molecules ... and Back Again? Cryst. GrowthDes. 7 (1), 4–9.

10.7423/XJENZA.2015.1.02 www.xjenza.org

Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.03

Research Article

Real-Time Modelling and Interpolation of Spatio-Temporal MarinePollution

N. Formosa, K. ScerriUniversity of Malta, Faculty of Engineering

Abstract. Due to the complexity of the interactionsinvolved in various dynamic systems, known physical,biological or chemical laws cannot adequately describethe dynamics behind these processes. The study of thesesystems thus depends on measurements often taken atvarious discrete spatial locations through time by noisysensors. For this reason, scientists often necessitate in-terpolative, visualisation and analytical tools to dealwith the large volumes of data common to these sys-tems. The starting point of this study is the seminalresearch by C. Shannon on sampling and reconstructiontheory and its various extensions. Based on recent workon the reconstruction of stochastic processes, this paperdevelops a novel real-time estimation method for non-stationary stochastic spatio-temporal behaviour basedon the Integro-Difference Equation (IDE). This meth-odology is applied to collected marine pollution datafrom a Norwegian fjord. Comparison of the results ob-tained by the proposed method with interpolators fromstate-of-the-art Geographical Information System (GIS)packages will show, that significantly superior results areobtained by including the temporal evolution in the spa-tial interpolations.

1 Introduction

Complex spatio-temporal interactions are exhibited invarious natural systems as witnessed in ecology, meteor-ology, physics and epidemiology. Such natural systemsevolve on a continuous spatial domain, but the datasets collected from these systems are made up of spa-tially localised measurements taken at different discretespatial locations through time by noisy sensors. Never-theless, for analysis and visualisation purposes, the un-derlying, spatially continuous processes must often bereconstructed from these discrete measurements.

Traditionally, spatial interactions have been math-ematically explained using spatial models, and severalinterpolation techniques are readily available. Never-theless, these methods provide interpolations that ig-nore the temporal behaviour of the process. Typicalexamples include Inverse Distance Weighting (IDW)(Poshtmasari, Sarvestani, Kamkar, Shataei & Sade-ghi, 2012), Gaussian Radial Basis Functions (GRBF)(Poshtmasari et al., 2012) and Kriging (Poshtmasari etal., 2012; Schabenberger & Gotway, 2005). However,since spatial measurements are also often repeatedlytaken in time, any interpolated field could greatly bene-fit from augmenting the temporal evolution of the datawith the spatial information. A modern approach to theinterpolation from spatial-temporal information may bebased on the reconstruction step of sampling theoryfirst attributed to C. Shannon’s seminal work (Shannon,1949). The aim of this paper is thus to provide a novelreal-time procedure for the estimation of the underlyingstochastic process from noisy observations that serves asa new tool for better interpolations. Another significantcontribution is the validation of the developed methodson a pollution data set highlighting the applicability ofthis methodology to real-world applications.

The remainder of this paper is structured as follows.The historical developments of sampling and reconstruc-tion theory fundamental to this research are presentedin Section II. More recent developments required for thiswork together with the chosen real-time estimation ap-proach are presented in Section III. Section IV is dedic-ated to an illustrative example highlighting the benefitsof this procedure. Finally, some concluding remarks aredrawn together with an overview of some possible futureadditions in Section V.

Correspondence to: N. Formosa (nicolette.formosa@um.edu.mt)

c© 2015 Xjenza Online

Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30. 24

2 Historical DevelopmentsThe starting point of this research is the seminal studyby C. Shannon on sampling and reconstruction theoryand its various extensions. Shannon first presented thefundamental framework to represent a bandlimited, con-tinuous time signal as a discrete sequence without anyloss of information in 1949 (Shannon, 1949). He alsointroduced an approach for a perfect reconstruction ofthe original time signal from its sampled representa-tion. Over the past decades, Shannon’s work has beenextended by various notable contributors as summar-ised by M. Unser in (Unser, 2000). These extensionsaim to relax some of Shannon’s assumptions and ex-tend the framework to account for non-ideal sampling(S. Ramani D. Van De Ville & Unser, 2008), multidi-mensional settings (Petersen & Middleton, 1962; Izen,2005; Feuer, 2004; Jr. & Sangsari, 1989) and more con-venient reconstruction spaces (Yao, 1967; Eldar & Un-ser, 2006; Eldar & Werther, 2005; Nashed & Walter,1991). Table 1 summarises the main contributions tosampling and reconstruction theory published since theseminal work of Shannon together with the contribu-tions envisaged through this work.

The restriction to unidimensional processes assumedin Shannon’s sampling theory prevents the applica-tion of this seminal theory to multidimensional signals.Petersen et. al. (Petersen & Middleton, 1964) extendedthis theory to accommodate other areas where physicalphenomena need to be measured in a multidimensionalcontinuum. Some conspicuous fields within this area in-clude meteorology, oceanography, seismology, acoustics,optics and radar (Petersen & Middleton, 1962). Pa-poulis (Papoulis, 1977) continued Shannon’s work stat-ing that it is unrealistic to assume that the acquisitiondevice is ideal. The theory thus developed shows theintroduction of a non-ideal acquisition device prior tosampling. However due to the introduction of a non-ideal acquisition device, a discrete correction filter priorto the reconstruction stage is required to limit its effect.

Moreover, sampling, as presented by Shannon is re-stricted to bandlimited signals. However, for non-bandlimited signals, Aldroubi and Unser (Aldroubi, Un-ser & Aldroubi, 1994) proposed the use of alternativereconstruction spaces which take into account both theinput signal and the sampling process. Conventionalsampling and interpolation theories provide solutions tothe problem of reconstructing a signal from its sampleslimited to a noise-free scenario. Perfect reconstructioncan, therefore, be achieved as long as this noise free ac-quisition is respected. In any real world application, themeasured signals are always corrupted by various de-grees of noise levels and, therefore, perfect reconstruc-tion is no longer attainable. Eldar and Unser in (Eldar &Unser, 2006), therefore, proposed the use of the Weiner

filter as the discrete correction filter to limit the effectof measurement noise on the reconstruction process.

Classical sampling theory assumes that each realisa-tion of the sampling processes is independent of allpast signals. Wikle showed that although such an as-sumption is valid in various temporal or spatial applic-ations, it is violated for spatio-temporal signals wheresuccessive spatial samples are temporally related tothe dynamics of the process under observation (Wikle,2002). Thus, sampling and reconstruction theory ofnon-stationary stochastic processes utilises a dynamicfilter to capture the temporal evolution of the process(Scerri, Dewar & Kadirkamanathan, 2008).

Building on the extensions discussed in this section,this work provides a method for the reconstruction ofsampled non-stationary stochastic processes observedin the presence of noise. A modern shift-invariant ap-proach is also adopted in a multi-dimensional domain.The sampling and reconstruction methods developed inthis paper are based on the same building blocks aspresented in (Scerri et al., 2008). Nevertheless, this re-search extends the work in (Scerri et al., 2008) such thatthe estimation of the kernel from noisy observations isimplemented in real-time. As opposed to previous work,the methods developed were validated on collected mar-ine pollution measurements.

3 Theoretical Developments

The temporally evolving signal z(s, t) ∈ L2 ∀ t whereL2 is the space of measurable, square-integrable, real-valued functions with s ∈ Rn and t ∈ Z+. The signalz(s, t) is continuous in the spatial domain (s) but dis-crete in the temporal domain (t).

As done in (Scerri et al., 2008), an approximation tothe spatially continuous signal z(s, t) in the shift invari-ant space V , is given by this projection into V given by

z(s, t) ≈∑j∈Z〈z(s, t), φxj (s)〉φxj (s) = x(t)> φx(s) (1)

where

x(k, t) = [〈z(s, t), φx1(s)〉 . . . 〈x(s, t), φxnx (s)〉]>

φi(s) = [φi1(s) . . . φini (s)]> (2)

φi(s) is some chosen reconstruction function. In thiswork it is assumed that the temporal evolution of thestochastic process z(s, t) can be described by the lin-ear spatial convolution of the previous field with an un-known kernel k(s) and subject to some additive noiseprocess given by:

z(s, t) =

∫z(., t− 1)k(.) + η(s, t) (3)

10.7423/XJENZA.2015.1.03 www.xjenza.org

25 Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30.

Table 1: Comparison of the various extensions to Shannon’s sampling theorem.

Method Dimensions AcquisitionDevice

ReconstructionSpace

Dynamics Noise Real-Time

(Shannon, 1949) 1-D Ideal Bandlimited No No No(Petersen & Middleton, 1964) Multi-D Ideal Bandlimited No No No(Papoulis, 1977) 1-D Non-ideal Bandlimited No No No(Aldroubi, Unser & Aldroubi, 1994) 1-D Non-ideal Shift-

InvariantNo No No

(Wikle, 2002) 1-D Non-ideal Harmonic Yes No No(Eldar & Unser, 2006) 1-D Non-ideal Shift-

invariantNo Yes No

(Scerri, Dewar & Kadirkamanathan, 2008) Multi-D Non-ideal Shift-invariant

Yes Yes No

This research (2014) Multi-D Non-ideal Shift-invariant

Yes Yes Yes

where η(s, t) is a zero mean stationary Gaussian noiseprocess with covariance Ση given by

Ση = COV[z(s, t), z(s + s, t+ t)] =

λ(s) if t = 0

0 otherwise

(4)Sampling is performed where filtering by the non-ideal

acquisition device is followed by an ideal sampler. Theoutput of the non-ideal filter is dictated by its impulseresponse, h(s), and is given by:

y(si, t) =

∫Sh(si − r)z(r, t)d r+v(t) (5)

where v(t) is a zero mean white Gaussian noise pro-cess uncorrelated with η(s, t) representing measurementnoise.

A major difficulty in making use of the unknown tem-poral dynamics in the reconstruction of z(s, t) is the con-tinuous nature of the convolution describing the tem-poral dynamics as given by (3). An approximate dis-cretised representation of the dynamics (3) and the ob-servation process (5) can be obtained by considering theapproximations:

z(s, t) ≈nx∑j=1

〈z(s, t), φxj (s)〉φxj (s) = x(t)> φx(s) (6)

k(s) ≈nθ∑j=1

〈k(s), φθj (s)〉φθj (s) = θ> φθ(s) (7)

h(s) ≈nϑ∑j=1

〈h(s), φϑj (s)〉φϑj (s) = ϑ> φϑ(s) (8)

λ(s) ≈n%∑j=1

〈λ(s), φ%j (s)〉φ%j (s) = %> φ%(s) (9)

where

x(t) = [〈z(s, t), φx1(s)〉 . . . 〈x(s, t), φxnx (s)〉]>

θ = [〈k(s), φθ1(s)〉 . . . 〈k(s), φθnθ (s)〉]>

ϑ = [〈h(s), φϑ1(s)〉 . . . 〈h(s), φϑnϑ (s)〉]>

% = [〈λ, (s)φ%1(s)〉 . . . 〈λ(s), φ%n% (s)〉]>

φi(s) = [φi1(s) . . . φini (s)]>

(10)

Based on these decompositions an approximate discreterepresentation of the dynamics (3) and the non-idealsampler (5) with known error bounds is given by The-orem 1 (Scerri et al., 2008).

Theorem 1. Using the spatially discrete representa-tions (6) to (9), the dynamic equation (3) and the ob-servation equation (5) can be approximated by the finitedimension state-space model

x(t+ 1) = A(θ)x(t) + w(t) (11)

and

y(t) = C(ϑ)x(t) + v(t) (12)

where

A(θ) = Ψ -1

∫Sφx(s)θ> Ξθ(s)ds (13)

Ψ =

∫Sφx(s)φx(s)>ds (14)

Ξθ(s) =

∫Sφθ(s− r)φx(r)>d r (15)

C(ϑ) =

ϑ> Ξϑ(s1)...

ϑ> Ξϑ(sny )

10.7423/XJENZA.2015.1.03 www.xjenza.org

Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30. 26

Ξϑ(s) =

∫Sφϑ(s− r)φx(r)>d r (16)

w(t) ∼ N (0,Σw) (17)

with

Σw = Ψ−1∫Sφ%(s)%(s)>Ξ%(s)dsΨ

−> (18)

Ξ%(s) =

∫Sφ%(s− r)φx(r)>d r (19)

and v(t) ∼ N (0,Σv) with Σv = σvIny ; with errors inthe approximation of z(s, t) given by

εz = |z(s, t)− x(t)> φx(s)| 6 ε′z

∫Rn:ν>νc

Φx(ν)dν

(20)where

ε′z = supRn:ν>νc

|Z(ν)Φ−1x (ν)| (21)

Proof is found in (Scerri et al., 2008). Since thedynamics are unknown, a joint estimation techniqueis being proposed to simultaneously reconstruct boththe stochastic non-stationary spatio-temporal processand the systems dynamics from the noise-corrupted ob-servations. This research necessitates real-time learn-ing of the model parameters and states, with the mostwidely used techniques being the online version of theExpectation-Maximisation (EM) algorithm, particle fil-ters and the Dual Kalman filter. Despite being a math-ematically rigorous technique, the online version of theEM algorithm is not easily applicable to different solu-tions due to the possible intractability of the maximisa-tion function (Ozkan, Fritsche & Gustafsson, 2012). Onthe other hand, although particle filters are easily applic-able to varied dynamics and provide accurate and rig-orous solutions, they carry a significant computationalburden which can easily become excessive in a spatio-temporal setting (Kwok, Fox & Meila, 2004). Thus, thedual Kalman filter was chosen for this application dueto its low computational cost, easy of adoption, rapidconvergence and well-studied behaviour (Wan, van derMerwe & Nelson, 1999).

The state estimates based upon the initial estimates ofθ are obtained from the first state space representationgiven in (3) and (12). Estimates of the parameter θbased on the values from first state-space model, arethen obtained from the state space representation givenby:

θ(t+ 1) = Iθ(t) + q(t) (22)

x(t+ 1) = B(x(t))θ + w(t) (23)

Note that (23) is a rewriting of (11) as done in (Scerriet al., 2008). This completes one run of the algorithm.Therefore, the use of dual Kalman filtering based on thestate space model describing both the states and para-meter evolution in real-time leads to the reconstructionof the spatio-temporal process z(s, t). Note that, thesensor’s spatial characteristics are assumed to be knownwhen the spatio-temporal process and the model dy-namics are being inferred. Such an assumption is reas-onable since sensor characteristics may be either givenby the manufacturer or measured by experimentation.Nevertheless, this assumption may be relaxed by addingthe estimation of the sensors parameters, ϑ, to the dualestimation procedure.

4 Marine Pollution Spread Mod-elling and Interpolation

An important contribution of this research was to applythe developed models on a real data set. To the author’sbest knowledge, this is the first application of such amethodology to real data. The marine pollution datasetwas provided by NORUS, a north American and Norwe-gian educational programme (Medina, Moline, Clark &Wood, 2011). Its main focus is the study of the effect ofclimate change on the ecosystems and living organisms.Part of the data required for NORUS is obtained viathe Slocum Glider (Medina et al., 2011), an autonom-ous underwater robot.

The dataset used for this project was taken fromthis glider deployed in the fjord of Svalbard, Norway.The glider’s mission extended from longitude 13.3042

East to 16.6875 East and latitude 78.1042 North and78.7042 North (Medina et al., 2011). During the 17-day mission, which began on June 30, 2009, the glidercollected a large data set of measurements for each ofits 20 onboard sensors. The data set available for thisresearch includes the position of the glider and the oxy-gen concentration at the robot’s location and the timeof measurement. Oxygen concentration is an importantindicator of water quality, and it is envisaged that thismethodology can adequately capture its spatio-temporalevolution from the measured data in real time. Such be-haviour is similar to other spreading pollutants in largebodies of water which may necessitate real-time spatio-temporal interpolation and prediction in the natural dis-aster scenarios.

4.1 Pre-Processing of Data

The stochastic spatio-temporal measurement of interestin this research is the oxygen concentration. Thus,these measurements were plotted against their discretesampling instances as shown in Figure 1. This plotshows variations in oxygen concentrations with time asthe glider is moving and collecting the data. It indic-

10.7423/XJENZA.2015.1.03 www.xjenza.org

27 Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30.

ates that there exists a cyclical variation in the oxygenreadings.

Figure 1: Oxygen concentration plotted against time.

With the aim to visualise the main causes of the cyc-lic variations in oxygen concentration, a similar graph ofthe depth against the time instances is shown in Figure2 . Both Figures 1 and 2 are seen to exhibit very sim-ilar behaviour, each having similar cycles at the sametime instances. This highlights the strong relationshipbetween oxygen concentration and depth.

Figure 2: Depth variation as the glider is moving in time.

A detrending procedure was carried out to capturethe apparent relationship between the oxygen level andthe depth. Detrending is often applied to remove a fea-ture thought to distort or obscure the relationships ofinterest (Chatfield, 2004). After detrending, to facil-itate interpretation and modelling of the dataset, trendestimation was carried out. In trend estimation, a modelis developed to pick up the main characteristics of thedata. Based on this pre-processing procedure, the re-siduals were clustered into four groups, with models foreach cluster given by:

• Cluster 1

l(t) = m1 + c1d+ r(t) (24)

• Cluster 2

l(t) = m2 + c2d+ r(t) (25)

• Cluster 3l(t) = m3 + r(t) (26)

• Cluster 4l(t) = m4 + r(t) (27)

where l(t) is the oxygen level, d is the depth, m is themean of the data, c is the linear regression model para-meter and r(t) represents the residual left after detrend-ing. Using Least Squares estimation, the model para-meters obtained are presented in Table 2.

Table 2: Simulation function and parameters.

Parameter Value

m1 3.1947c1 1.2115m2 −1.1732c2 −0.9596m3 1.0678m4 −8.6269

A plot of the detrended version of the original dataset is given in Figure 3. This figure indicates that theresiduals form a horizontal band about zero and have avery small variance up to a depth of 6m. Moving, how-ever, to a greater depth, even though the mean remainsaround zero, the variance is significantly larger.

Figure 3: Detrended Data.

10.7423/XJENZA.2015.1.03 www.xjenza.org

Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30. 28

To verify if significant spatial and temporal correla-tions are still present after detrending, correlation coef-ficients are obtained. The correlogram presented in Fig-ure 4, facilitates the interpretation of the results of theautocorrelation coefficients. The 95% confidence inter-vals are also included in Figure 4. Values outside theseconfidence intervals are considered to be statistically in-significant. Figure 4 shows that significant spatial andtemporal correlations still exist in the residues since val-ues at low lag exhibit correlation coefficients well beyondthe 95% confidence interval.

Figure 4: Correlogram for the Detrended Data.

4.2 Spatio-Temporal Modelling

Since the pollution spread dynamics are unknown, thereconstruction being proposed reconstructs both thespatio-temporal process and the model kernel via thejoint estimation technique in real-time. This is carriedout based on the spatial location of the measurementsand their evolution through time. Fourier and correla-tion analysis were used to identify the bandwidth andkernel support of the data as suggested in (Scerri et al.,2008). Based on this analysis, a spatio-temporal modelwas proposed using cubic B-splines as the basis func-tions. The functions and parameters for reconstructionof both the process and the underlying dynamics arepresented in Table 3.

The tests carried out on the proposed interpolationtechnique measure the accuracy of the spatio-temporalinterpolation. To quantify the accuracy of the recon-structed spatio-temporal process, Table 4 shows theRoot Mean Square Error (RMSE) difference between themeasured spatio-temporal process and the reconstruc-ted process. These RMSE measures indicate relativelysmall errors between the measurements and the recon-

Table 3: Simulation function and parameters.

Function and Parameter Simulation Value

λ(s) = δ(s)

h(s) = 0.3 exp( s2

3 )ny = 3134nθ = 3nϑ = 3

nx = nρ = 7basis locations for z(s, t) = −11,−7.5, ..., 10basis locations for k(s) = −3.5, 0, 3.5basis locations for h(s) = −3.5, 0, 3.5

structed field. The values from Table 4 show that forthe randomly chosen time points, the proposed method-ology for real-time estimation of the parameters and thestates obtained a reasonably smoothed representationof the measured data. Running on standard computerhardware with Intel(R) Core(TM) i7-2670QM CPU at2.20GHz processor and 8GB of main memory, the com-putational time for each time point was approximately0.388s, thus near immediate estimates are obtained foreach now measured field. Most importantly, the overallRMSE over all time points is measured at 0.5345mg/l,indicating that the proposed procedure for real-timeparameter and state estimation was able to pick themain process behaviour at each time and the interpol-ated behaviour is a good representative sample of thegeneral behaviour.

The widely used 10-fold validation method (McLach-lan, Do & Ambroise, 2004) is used as a validation ofthe proposed methodology. Thus, a sample of 300-timepoints is removed from the data set for each run to beused as a validation set. The advantage of this methodis that all observations are used for both training andvalidation, and each observation is used for validationonly once. The RMSE of each fold used as a validationset is presented in Table 5. For each fold, the resultsshow that good estimates are obtained with low RMSEsand relatively low percentage errors. Moreover for eachfold tested, the kernel parameters converged to the samevalues, obtaining a repeatable kernel reconstruction.

4.3 Comparison with Other Interpola-tion Techniques

Furthermore, standard spatial interpolation techniquesprovided by the latest GIS software were used to com-pare their performance with the results obtained fromthe proposed real-time methodology. 10-fold validationwas carried out for each interpolation technique avail-able in GIS software. The results from each interpol-ation technique are included in Table 6 together withthe spatio-temporal methodology developed in this pa-

10.7423/XJENZA.2015.1.03 www.xjenza.org

29 Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30.

Table 4: RMSE of the reconstructed spatio-temporal process for the randomly chosen time points.

t RMSE (mg/l) % error t RMSE (mg/l) % error

89 0.5004 0.4924 162 2.0910 1.9366332 0.5366 0.4910 520 1.2928 1.1600787 0.2517 0.2444 890 2.1250 2.7420

1222 0.2904 0.2851 1329 1.2021 1.75631546 0.3534 0.3440 1800 0.5198 0.50721844 0.3303 0.3125 1963 0.2165 0.20902385 1.0642 1.0048 2855 2.6440 2.13413005 0.3193 0.3135 3118 0.3247 0.3188

Table 5: Overall RMSE of the reconstructed spatio-temporal process within each fold.

t RMSE (mg/l) % error

1:300 0.2054 0.2015301:600 0.2942 0.2797601:900 0.3711 0.3506901:1200 0.4049 0.3826

1201:1500 0.2956 0.27861501:1800 0.2396 0.22601801:2100 0.3085 0.29062101:2400 0.1649 0.15552401:2700 0.2205 0.20822701:3000 0.2262 0.2133

per. These results show that the lowest RMSE from allstandard interpolation techniques is obtained via empir-ical Bayesian kriging. Nevertheless, this is still signific-antly outperformed by the spatio-temporal methodologydeveloped in this paper.

Table 6: Summary of the results for all the interpolatorswhich were considered.

All Functions in GIS software and de-veloped models

RMSE(mg/l)

Inverse Distance Weighting 1.0521Global Polynomial Interpolation - Order10

2.1503

Radial Basis Functions - MultiquadraticKernel

1.0156

Local Polynomial Interpolator - Order 1with Exponential Kernel

1.7826

Kriging/CoKriging - Ordinary Kriging 1.0283Kriging/CoKriging - Universal Kriging 1.0283Empirical Bayesian Kriging 0.9862Diffusion Kernel 2.0332Spatio-Temporal Modelling 0.2731

5 ConclusionThis research presented a structured approach to data-driven spatio-temporal modelling and spatial interpol-ation utilising a state space representation of the IDEwith parameter and state estimation in real-time. Clas-sical interpolation theory assumes that the data col-lected from sensors is independent of all past signals.However, in spatio-temporal signals, successive spatialsamples are temporally related based on the dynam-ics of the process under observation. Novel methodsfor modelling and spatial interpolation were developedbased on multidimensional sampling theory, to caterfor noise-corrupted observations from spatially discretesensors. A real-time estimation procedure was utilisedto estimate the model dynamics and the hidden spatio-temporal process with the use of the dual Kalman filter.This methodology was applied to a real data set thusproviding the very first validation of these methods ongathered, rather than simulated data. The proposedmodels together with the methodology developed, ob-tained better interpolative results than any competitivespatial interpolator. This can be attributed to the use ofthe temporal information in the developed interpolator.Thus, these spatio-temporal models may aid in both thevisualisations of the data and a better understanding ofthe evolving process.

To further enhance the applicability of the proposedmethodology, various other extensions could be con-sidered. In one such extension, the assumptions of fullknowledge of the spatial response of the sensor and thesupport of the continuous functions of the IDE is re-moved. Furthermore, the homogeneous representationof the IDE can be altered to a spatially heterogeneousrepresentation allowing for the modelling and interpol-ation of more complex dynamics.

References

Aldroubi, A., Unser, M. & Aldroubi, A. (1994).Sampling procedures in function spaces andasymptotic equivalence with shannon’s sampling

10.7423/XJENZA.2015.1.03 www.xjenza.org

Formosa, N. and Scerri, K. (2015).Xjenza Online, 3:23–30. 30

theory. Numer. Funct. Anal. Optim. 15 (1-2), 1–21.

Chatfield, C. (2004). The Analysis of Time Series: AnIntoduction (6th). CRC Press.

Eldar, Y. C. & Unser, M. (2006). Nonideal Sampling andInterpolation from Noisy Observations in Shift- Invariant Spaces. IEEE Trans. Signal Process.54 (7), 2636–2650.

Eldar, Y. C. & Werther, T. (2005). General Frameworkfor Consistent Sampling in Hilbert Spaces. Int. J.Wavelets, Multiresolution Inf. Process, 3 (4), 497–509.

Feuer, A. (2004). On the Necessity of Papoulis’ Res-ult for Multidimensional GSE. IEEE Trans. SignalProcess. 11 (4), 420–422.

Izen, S. H. (2005). Generalised Sampling Expansionon Lattices. IEEE Trans. Signal Process. 53 (6),1949–1963.

Jr., J. L. B. & Sangsari, K. (1989). Sampling Recon-struction of N-Dimensional Band-Limited Imagesafter Multilinear Filtering. IEEE Trans. CircuitsSyst. 36 (7), 1035–1038.

Kwok, C., Fox, D. & Meila, M. (2004). Real-TimeParticle Filters. Proc. IEEE, 92 (3), 469–484.

McLachlan, G. J., Do, K.-A. & Ambroise, C.(2004). Analyzing Microarray Gene ExpressionData. Wiley Series in Probability and Statistics.Hoboken, NJ, USA: John Wiley & Sons, Inc.

Medina, D., Moline, M., Clark, C. & Wood, J. Z. (2011).Real - Time Visualizations of Ocean Data Col-lected by the NORUS Glider in Svalbard, Norway(Master’s thesis, California Polytechnic State Uni-versity).

Nashed, M. Z. & Walter, G. G. (1991). GeneralSampling Theorems for Functions in ReproducingKernel Hilbert Spaces. Math. Control Signals Syst.4, 373–412.

Ozkan, E., Fritsche, C. & Gustafsson, F. (2012). OnlineEM Algorithm for Joint State and Mixture Meas-urement Noise Estimation. Proc. IEEE Int. Conf.Inf. Fusion, 1935–1940.

Papoulis, A. (1977). Generalized Sampling Expansion.IEEE Trans. Circuit Syst. 24, 652–654.

Petersen, D. P. & Middleton, D. (1962). Sampling andReconstruction of Wave - Number - Limited Func-tions in N - Dimensonal Euclidean Spaces. Inf.Control, 5, 279–323.

Petersen, D. P. & Middleton, D. (1964). Reconstructionof Multidimensional Stochastic Fields from Dis-crete Measurements of Amplitude and Gradient.Inf. Control, 7, 445–476.

Poshtmasari, H. K., Sarvestani, Z. T., Kamkar, B.,Shataei, S. & Sadeghi, S. (2012). Comparison ofInterpolation Methods for Estimating pH and ECin Agricultural Fields of Golestan Province. Int.J. Agric. Crop Sci. 4 (4), 157–167.

S. Ramani D. Van De Ville, T. B. & Unser, M. (2008).Nonideal Sampling and Regularization Theory.IEEE Trans. Signal Process. 56 (3), 1055–1070.

Scerri, K., Dewar, M. & Kadirkamanathan, V. (2008).Estimation and Model Selection for an IDE-basedSpatio-Temporal Model. IEEE Trans. Signal Pro-cess. 57 (2), 482–492.

Schabenberger, O. & Gotway, C. A. (2005). Statist-ical Methods for Spatial Data Analysis (1st). CRCPress.

Shannon, C. E. (1949). Communication in the Presenceof Noise. 37, 10–21.

Unser, M. (2000). Sampling - 50 Years After Shannon.IEEE Proc. 88 (4), 569–587.

Wan, E., van der Merwe & Nelson, A. (1999). Dual Es-timation and the Unscented Transformation. In S.Solla, T. Leen & K.-R. Muller (Eds.), Proc. neuralinf. process. syst. conf. (pp. 666–672). MIT Press.

Wikle, C. K. (2002). A kernel-based spectral modelfor non-Gaussian spatio-temporal processes. Stat.Modelling, 2 (4), 299–314.

Yao, K. (1967). Applications of Reproducing Kernel Hil-bert Spaces - Band-Limited Models. Inf. Control,11, 429–444.

10.7423/XJENZA.2015.1.03 www.xjenza.org

Tanti, E. and Francalanza, A. (2015).Xjenza Online, 3:3135.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.04

Research Article

Towards Sound Refactoring in Erlang

E. Tanti, A. Francalanza

Department of Computer Science, Faculty of ICT, University of Malta

Abstract. Erlang is an actor-based programminglanguage used extensively for building concurrent, react-ive systems that are highly available and suer minimumdowntime. Such systems are often mission critical, mak-ing system correctness vital. Refactoring is code re-structuring that improves the code but does not changebehaviour. While using automated refactoring tools isless error-prone than performing refactorings manually,automated refactoring tools still cannot guarantee thatthe refactoring is correct, i.e., program behaviour is pre-served. This leads to lack of trust in automated refact-oring tools. We rst survey solutions to this problemproposed in the literature. Erlang refactoring tools ascommonly use approximation techniques which do notguarantee behaviour while some other works propose theuse of formal methodologies. In this work we aim todevelop a formal methodology for refactoring Erlang

code. We study behavioural preorders, with a special fo-cus on the testing preorder as it seems most suited toour purpose.

1 Introduction

There exist a number of tools which assist with theconstruction and maintenance of software. One ex-ample of such tools are refactoring tools which auto-mate the process of modifying a software system in sucha way that improves the internal structure of programcode but at the same time does not alter the externalfunctionality or behaviour of the code. Examples ofsuch improvements include improved readability, suchas through compliance with certain code practices, im-proved eciency, reusability, extensibility and maintain-ability (Mens & Tourwé, 2004). Also, recently (Brown,Hammond, Danelutto & Kilpatrick, 2012, 2013), theuse of refactoring tools has been proposed to automat-ically preform the complex task of transforming singlethreaded code into concurrent code.

The growth of multicore, and even more recentlymany-core, computer processors has made concurrencya crucial consideration for developers. Taking advant-age of the potential speedup that can be achieved usingmulticore computers is one of the main driving forces forprogramming concurrently. However, concurrent pro-grams are dicult for developers to construct and fortools to analyse.There exist a number of concurrency models which al-

low developers to construct concurrent programs. Theactor model is one such abstraction which eases the con-struction of concurrent programs, providing an alternat-ive to the currently predominant shared memory concur-rency model. Actors are suitable for structuring highlyconcurrent software systems which scale up to many-core processors, as well as scale out to clusters and thecloud (Cesarini & Thompson, 2009; Haller & Sommers,2011). This has led to a steady stream of research andindustrial development, contributing to a renewed in-terest in actors in academia (Karmani, Shali & Agha,2009; Haller, 2012; Sutter & Larus, 2005). Examples ofprogramming languages which have adopted the actormodel include Erlang (Armstrong, 2007; Cesarini &Thompson, 2009) and Scala (Haller & Sommers, 2011).There also exist a number of actor frameworks such asAkka which run on the Java Virtual Machine (JVM)with APIs available for both Java and Scala (Haller,2012) amd Microsoft's Asynchronous Agents Library for.NET (Karmani et al., 2009). In our work we havechosen to focus on Erlang as it is one of the mostmature actor languages (Armstrong, 2010).There exist a number of refactoring tools for actor-

based programming languages such as Erlang, Ti-dier (Sagonas & Avgerinos, 2009) and Wrangler (Li,Thompson, Orosz & Tóth, 2008). The use of automatedrefactoring tools is generally faster and less likely to in-troduce mistakes than refactoring manually (Murphy-Hill, Parnin & Black, 2009). Despite this, refactor-

Correspondence to: E. Tanti (erica.tanti.09@um.edu.mt)

c© 2015 Xjenza Online

Tanti, E. and Francalanza, A. (2015).Xjenza Online, 3:3135. 32

ing tools are, as discussed in a survey by Murphy-Hill(Murphy-Hill, 2007), largely underused. Amongst thereasons for this phenomenon the people surveyed citedlack of trust in automated refactoring tools. To counterthis problem, most of the refactorings provided, includ-ing those by existing Erlang refactoring tools suchas Tidier (Sagonas & Avgerinos, 2009), are simple asthis is the only way to ensure that behaviour is pre-served. Despite this, bugs may still be found in refactor-ing tools, even after extensive testing (Li & Thompson,2008; Soares, Gheyi, Serey & Massoni, 2010). There-fore tools permitting more complex refactorings such asWrangler (Li et al., 2008) use side-conditions or asser-tions to attempt to ascertain whether a refactoring isvalid, but do not guarantee behaviour preservation. Tomitigate the problem, such tools, including Wrangler,are often semi-automatic, thus necessitating some hu-man interaction and support an undo mechanism whichMens and Tourwé (2004) state is required when a toolcannot guarantee behaviour preservation.In this paper, we rst survey and discuss existing

methodologies for determining behaviour preservationin refactoring tools in Section 2. In this section, we dis-cuss both approximation techniques and precise formaltechniques. We then discuss in Section 3 how to applyprevious work to develop a formal behavioural theoryfor Erlang and which may be applied to Erlang re-factoring tools. Section 4 concludes.

2 Techniques used to prevent er-

roneous refactorings

In this section we give an overview of techniques com-monly used to prevent erroneous refactorings. We rstpresent a brief overview of approximation techniquescommonly used which do not guarantee behaviour pre-servation. We then discuss formal methodologies usedin previous work.

2.1 Approximation Techniques

A number of approximation techniques exist for determ-ining behaviour preservation. One relatively simple andcommon way to check for a refactoring's correctness isto perform conditional checks on the code. These comein the form of (1) invariants that should remain satisedthroughout the process, and (2) pre- and postconditions

that should hold before and after the refactoring hasbeen applied respectively. These conditions serve onlyto approximate correctness as only some of the set of allpossible properties are checked. Additionally, staticallychecking certain properties may be computationally ex-pensive or impossible. An example of such work includesLi and Thompson (2007) work checking Wrangler (Li etal., 2008) refactorings and Drienyovszky et al.'s work(Drienyovszky, Horpácsi & Thompson, 2010) checking

Wrangler and RefactorErl refactorings, using Quviq'sautomated testing tool QuickCheck (Hughes, 2007).However such methods are only approximations and

thus do not supply us with denitive guarantees. Insafety-critical software, undoing such a refactoring whena bug is found, and thus when the damage may alreadyhave been done, is unsatisfactory. The expense of in-troducing a bug, even for a short time, outweighs thepossible benets of refactoring.

2.2 Formal Methodologies

The problem of introducing bugs despite the precau-tions described above could potentially be avoided ifthere was a way to truly ensure that the refactoringis correct, i.e., that behaviour has been preserved. Tothis end, there has been some previous work on the useof formal techniques which may be used to ensure be-haviour preservation. Work carried out by Ward andBennett (Ward & Bennett, 1995), Li and Thompson(Li & Thompson, 2005) and Sultana and Thompson(Sultana & Thompson, 2008) use a number of dier-ent formal techniques to determine whether behaviourhas been preserved, similar to the goals of this work.Complementary to such techniques there are also graphtransformations whereby software is represented as agraph, and restructurings correspond to transformationrules. Graph transformations are ideal for replacingoccurrences of poor design patterns with good designpatterns (Mens, Demeyer, Bois, Stenten & Gorp, 2003;Mens & Tourwé, 2004). These graph transformationsmust be formally proven to be correct.

2.2.1 Previous applications of formal methodo-

logies

Ward and Bennett's (Ward & Bennett, 1995) WideSpectrum Language (WSL) is a simple, unambiguous,formal language with pre-proven refactorings, thus guar-anteeing the refactoring's correctness. Translators arethen written to and from WSL to the target language.Using an intermediate language in this tool proved tobe useful beyond the usual advantages of code improve-ment as it may be used for reverse engineering, thatis to generate specications and to translate code fromone language to another. However, as their languageis very general purpose, their translated code is oftencumbersome and the translations themselves cannot beformally proven to be accurate.Work (Li & Thompson, 2005) exists on proving be-

haviour preserving refactorings for the Haskell, which isa functional programming language like Erlang. Theyformalise a core subset of Haskell using a simple lambda-calculus and a number of semantic equivalence reductionrules. Each rule is known to preserve semantic equival-ence, thus a refactoring made up of a combination ofthese rules is also considered semantically equivalent.

10.7423/XJENZA.2015.1.04 www.xjenza.org

33 Tanti, E. and Francalanza, A. (2015).Xjenza Online, 3:3135.

However, they do not specify the equivalence: theoryused (Li & Thompson, 2005).More recently, and more similar to our ultimate goal,

is the work of Sultana and Thompson (2008) who usea proof assistant Isabelle/HOL to verify refactorings fortwo fragments (Sultana & Thompson, 2008) of the pro-gramming languages Haskell and Erlang, using pre-conditions and behavioural equivalence. They aimed tostudy the methodology for equivalence checking, whichcould later be extended for larger, more realistic lan-guages and for larger refactorings. Their preliminaryobservations are promising.

3 Aims

Given the previous work on formal methodologiesfor guaranteeing behaviour preservation for refactoringtools, particularly Sultana and Thompson's (Sultana &Thompson, 2008) work, we felt that it would be ap-propriate to delve deeper into the topic of behaviouralequivalence. To this end, we rst discuss formal meansof modelling the Erlang language which may then beused to prove behavioural properties of the language

3.1 Language Formalisations

To be able to study behavioural equivalence theories forrefactoring Erlang code, we rst construct a formalcalculus for the language. In this section we give areview of existing actor and Erlang language calculiwhich may be used or adapted for this work.There is a series of works which build on each other

that attempt to formalise Erlang, namely those ofFredlund (2001), Claessen and Svensson (2005), Svens-son and Fredlund (2007) and (Svensson, Fredlund &Earle, 2010). These formalisations occurred post-hoc,that is after the Erlang language was created. Theiraim was to create a faithful representation of the Er-lang compiler so as to study the behaviour of actors inlocal and distributed settings and encourage discussionon how the Erlang language may be better understoodand improved. Francalanza and Seychell (2013) modelthe basic Erlang functions and also the tracing mech-anism available in the Erlang Virtual Machine (EVM).These previous works dier from ours as they do not dis-cuss behavioural theories.Similar to our goals, Thati (2003) and Gaspari and

Zavattaro (1999, 2000) develop actor calculi to studyequivalence theories. Thati develops an actor calculus,in this case to study the may testing preorder. How-ever, while his aim is closest to ours, the actor languagedeveloped in this work is essentially a variant of asyn-chronous π-calculus with a type system enforcing discip-line on the use of names. This means he forgoes actorfeatures such as mailboxes and the become primitive isreduced to an input (without pattern matching). Gas-

pari and Zavattaro note that a lot of research, includingequivalence theories, use process algebras, mathemat-ical languages used for describing and verifying proper-ties of concurrent systems such as CCS and π-calculus.However, they argue that such calculi are very dier-ent from the calculi developed for the actor model andreal object-oriented and actor languages. They there-fore develop an actor calculus and Labelled TransitionSystems (LTS) that have a clean formal denition, likeπ-calculus, while capturing essential features specic toactors. They then use the language to study asynchron-ous bisimulation equivalence.

3.2 Behavioural equivalences

Given a language formalisation for Erlang, we maythen study behavioural equivalences. Behavioural equi-valences are used to determine in which cases two sub-programs oer similar interaction capabilities when runin the context of any larger program, and thus seem likean ideal candidate to be applied to refactoring tools.However, there is a large number of relevant propertieswhich may need to be considered. This has led to thedevelopment of many dierent equivalences theories. Anextensive description and a lattice of known behaviouralequivalences and preorders over LTS, ordered by inclu-sion, can be found in the literature e.g. (van Glabbeek,2001). To determine which behavioural theory should beused, we must rst have a well-dened notion of whata safe or acceptable refactoring is. One denition, asdened in the original thesis about refactoring by Op-dyke (Opdyke, 1992), is that behaviour is preserved iffor the same sequence of inputs one receives the samesequence of outputs. However, this is not always su-cient, for example:

• Real-time software should preserve all temporalconstraints, and

• Embedded software has memory constraints andpower consumption which may need to be preservedby refactoring (Mens & Tourwé, 2004).

In our case we have decided to focus solely on thedenition of behaviour preservation provided by Opdyke(Opdyke, 1992) mentioned previously. This denitionis similar to that for the testing behavioural equival-ences by De Nicola and Hennessy (De Nicola & Hen-nessy, 1984). In their work De Nicola and Hennessystate that if given two programs, A and B, behaviour ispreserved if when A passes a set of tests, then B will alsopass these tests. When applied to refactoring A wouldbe the program before refactoring and B the programafter refactoring. By generalising the set of tests to allpossible tests, this condition may be reformulated as:

A <∼testB i for all T.(A passes T

)implies

(B passes T

)10.7423/XJENZA.2015.1.04 www.xjenza.org

Tanti, E. and Francalanza, A. (2015).Xjenza Online, 3:3135. 34

In this way, we will be able to determine that theoutput after refactoring has not changed and thus therefactoring is correct.Unfortunately this reasoning is impractical, as one

cannot run the set of all possible tests in a nite amountof time. We therefore plan to investigate an alternativetheory that facilitates reasoning about our testing equi-valence. The technical development does not follow dir-ectly from previous work (Castellani & Hennessy, 1998;Boreale, De Nicola & Pugliese, 2002; Thati, 2003) ascertain characteristics pertaining to actor systems com-plicate the development of our alternative preorder.

4 Conclusion

In this paper, we discussed the problem of ensuringbehaviour preservation of a program after refactor-ing. While a number of approximation techniques ex-ist, these cannot ensure that bugs are not introducedwhen refactoring. Thus, we focus on formal techniques,mainly behavioural equivalence, as a means of formallyproving and thus ensuring that program behaviour wasnot modied when refactoring.Using the literature surveyed as a starting point, in

our work we develop a formal calculus for the Erlanglanguage and a behavioural equivalence technique usingthe calculus which is suitable for refactoring tools. Onepossible starting point is using testing equivalences byDe Nicola and Hennessy (De Nicola & Hennessy, 1984)as it is very similar to the denition of behaviour pre-servation found in the original thesis about refactoringby Opdyke (Opdyke, 1992). As reasoning using testingequivalence is often cumbersome, we are also tasked withdeveloping an alternative theory for reasoning abouttesting equivalence, tailored for our Erlang calculus.

Acknowledgement

This study was carried out following the award ofa STEPS scholarship which is part-nanced by theEuropean Union European Social Fund (ESF) un-der Operational Programme II Cohesion Policy 2007-2013, Empowering People for More Jobs and a BetterQuality of Life.

References

Armstrong, J. (2007). Programming Erlang - Software

for a Concurrent World. The Pragmatic Book-shelf.

Armstrong, J. (2010). Erlang. Commun. ACM, 53 (9),6875.

Boreale, M., De Nicola, R. & Pugliese, R. (2002). Traceand Testing Equivalence on Asynchronous Pro-cesses. Inf. Comput. 172 (2), 139164.

Brown, C., Hammond, K., Danelutto, M. & Kilpatrick,P. (2012). A language-independent parallel refact-oring framework. In Proc. fth work. refactoring

tools (pp. 5458). WRT '12. New York, NY, USA:ACM.

Brown, C., Hammond, K., Danelutto, M., Kilpatrick, P.,Schöner, H. & Breddin, T. (2013). Paraphrasing:Generating Parallel Programs Using Refactoring.In B. Beckert, F. Damiani, F. S. Boer & M. M.Bonsangue (Eds.), Form. methods components ob-jects (Vol. 7542, pp. 237256). Lecture Notes inComputer Science. Springer Berlin Heidelberg.

Castellani, I. & Hennessy, M. (1998). Testing Theoriesfor Asynchronous Languages. In Fsttcs (pp. 90101).

Cesarini, F. & Thompson, S. (2009). ERLANG Pro-

gramming (Media, Inc). O'Reilly.Claessen, K. & Svensson, H. (2005). A semantics for

distributed Erlang. Proc. 2005 ACM SIGPLAN

Work. Erlang - ERLANG '05, 78.De Nicola, R. & Hennessy, M. (1984). Testing equival-

ences for processes. Theor. Comput. Sci. 34 (1-2),83133.

Drienyovszky, D., Horpácsi, D. & Thompson, S. (2010).Quickchecking refactoring tools. In Proc. 9th acm

sigplan work. erlang (pp. 7580). Erlang '10. NewYork, NY, USA: ACM.

Francalanza, A. & Seychell, A. (2013). Synthesising Cor-rect Concurrent Runtime Monitors - (ExtendedAbstract). In Rv (pp. 112129).

Fredlund, L.-Å. (2001). A Framework for Reasoning

about Erlang Code (Doctoral dissertation, RoyalInstitute of Technology, Stockholm, Sweden).

Gaspari, M. & Zavattaro, G. (1999). An Algebra of Act-ors. In P. Ciancarini, A. Fantechi & R. Gorrieri(Eds.), Form. methods open object-based distrib.

syst. (Vol. 10, pp. 318). IFIP - The InternationalFederation for Information Processing. SpringerUS.

Gaspari, M. & Zavattaro, G. (2000). An Actor Al-gebra for Specifying Distributed Systems: the Hur-ried Philosophers Case Study. In Concurr. object-

oriented program. petri nets, lect. notes comput.

sci. (pp. 428444). Springer-Verlag.Haller, P. (2012). On the Integration of the Actor Model

in Mainstream Technologies: The Scala Perspect-ive. In Proc. 2nd ed. program. syst. lang. appl.

based actors, agents, decentralized control abstr.

(pp. 16). AGERE! New York, NY, USA: ACM.Haller, P. & Sommers, F. (2011). Actors in Scala - Con-

current programming for the multi-core era. Ar-tima Press, California.

Hughes, J. (2007). QuickCheck testing for fun andprot. In Proc. 9th int. conf. pract. asp. declar.

lang. (pp. 132). PADL'07. Berlin, Heidelberg:Springer-Verlag.

10.7423/XJENZA.2015.1.04 www.xjenza.org

35 Tanti, E. and Francalanza, A. (2015).Xjenza Online, 3:3135.

Karmani, R. K., Shali, A. & Agha, G. (2009). ActorFrameworks for the JVM Platform: A Comparat-ive Analysis. In Proc. 7th int. conf. princ. pract.

program. java (pp. 1120). PPPJ '09. New York,NY, USA: ACM.

Li, H. & Thompson, S. (2005). Formalisation of Haskellrefactorings. Trends Funct. Program. 265280.

Li, H. & Thompson, S. (2007). Testing Erlang Refact-orings with QuickCheck. In 19th int. symp. imple-

ment. appl. funct. lang. i 2007, lncs. Freiburg,Germany.

Li, H. & Thompson, S. (2008). Testing Erlang Refact-orings with QuickCheck. In O. Chitil, Z. Horváth& V. Zsók (Eds.), Implement. appl. funct. lang.(Vol. 5083, pp. 1936). Lecture Notes in ComputerScience. Springer Berlin Heidelberg.

Li, H., Thompson, S., Orosz, G. & Tóth, M. (2008).Refactoring with Wrangler, updated: data andprocess refactorings, and integration with eclipse.Proc. 7th ACM SIGPLAN Work. ERLANG.

Mens, T., Demeyer, S., Bois, B. D., Stenten, H. & Gorp,P. V. (2003). Refactoring: Current research and fu-ture trends. In Proc. third work. lang. descr. tools

appl. (pp. 120130). Electronic.Mens, T. & Tourwé, T. (2004). A survey of software

refactoring. Softw. Eng. IEEE Trans. 30 (2), 126139.

Murphy-Hill, E. (2007). Programmer-Friendly Refactor-ing Tools. Other.

Murphy-Hill, E., Parnin, C. & Black, A. P. (2009). Howwe refactor, and how we know it. In Proc. 31st int.conf. softw. eng. (pp. 287297). ICSE '09. Wash-ington, DC, USA: IEEE Computer Society.

Opdyke, W. F. (1992). Refactoring Object-oriented

Frameworks (Doctoral dissertation, Champaign,IL, USA).

Sagonas, K. & Avgerinos, T. (2009). Automatic refact-oring of Erlang programs. Proc. 11th ACM SIG-

PLAN Conf. Princ. Pract. Declar. Program. -

PPDP '09, 13.Soares, G., Gheyi, R., Serey, D. & Massoni, T. (2010).

Making Program Refactoring Safer. IEEE Softw.

27 (4), 5257.Sultana, N. & Thompson, S. (2008). Mechanical veri-

cation of refactorings. Proc. 2008 ACM SIG-

PLAN Symp. Partial Eval. Semant. Progr. Manip.

- PEPM '08, 51.Sutter, H. & Larus, J. (2005). Software and the Concur-

rency Revolution. Queue, 3 (7), 5462.Svensson, H. & Fredlund, L.-Å. (2007). A more accurate

semantics for distributed erlang. Proc. 2007 SIG-

PLAN Work. Erlang Work. - Erlang '07, 43.Svensson, H., Fredlund, L.-Å. & Earle, C. B. (2010). A

unied semantics for future Erlang. In Proc. 9th

acm sigplan work. erlang (pp. 2332). ACM.Thati, P. (2003). A theory of testing for asynchronous

concurrent systems (Doctoral dissertation).van Glabbeek, R. J. (2001). The Linear Time-Branching

Time Spectrum I - The Semantics of Concrete,Sequential Processes. In Handb. process algebr.

chapter 1 (pp. 399). Elsevier.Ward, M. & Bennett, K. H. (1995). Formal methods to

aid the evolution of software. Int. J. Softw. Eng.Knowl. Eng. 5 (1), 118.

10.7423/XJENZA.2015.1.04 www.xjenza.org

Azzopardi, S., Buttigieg, E., and Licata, L. (2015).Xjenza Online, 3:36–40.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.05

Review Article

Symbiotics of History and Social Psychology Understanding SocialRepresentations of History in Europe

S. Azzopardi1, E. Buttigieg1, L. Licata21University of Malta, Msida MSD 2080, Malta2Universite Libre de Bruxelles (ULB), avenue Franklin D. Roosevelt 50, 1050 Brussels, Belgium

Abstract. COST Action IS1205 aims at advancingknowledge and promoting networking among histori-ans and social psychologists to analyse the role playedby social representations of history in Europe. Socialrepresentations of history are central to the identity ofgroups that may or may not form the majority in anygiven country. In Europe, these representations are atbest diverse, at worst fragmented, among various na-tional and ethnic groups, either in the same countryor across the continent. If left unexplored and unex-plained, these social (mis)representations can incite ad-verse emotions, in turn influencing group behavioursand possibly leading to intergroup rivalry. Bridging thetwo disciplines through representatives from 28 coun-tries, Action IS1205 addresses this issue by coordinat-ing research on the role of: social cognitive processes inshaping lay representations of history; lay representa-tion of history through the concepts of nationhood andidentities; social-psychological studies of the narrativetransmission of history through textbooks and the me-dia; lay representation of history and group-based emo-tions in shaping attitudes, intergroup conflict and re-conciliation processes.

1 Introduction

1.1 What is the relationship betweenHistory and Social Psychology?

The field of psychology is largely defined as ‘the scienceof human behaviour’. Social psychology is that branchwhich deals specifically with human interaction. It seeksto establish general laws which describe and explain so-cial relations. If such general principles of human be-haviour could be established, it becomes more possible

to determine social contexts in such a way as to offeroptimal benefits to members of society, predict socialbehaviour, and reduce conflict Gergen, 1973.

History as a discipline is concerned with the course ofinternational relations, nations and social institutions(political, financial, educational structures), economicand social development, social groups and movements(civil right protestors, student activists, strikers, tradeunions), and of groups of people (women, minorities,children, migrant labourers). Historians focus on thehistory of particular systems, analyse the history of peri-ods and events, and study the processes and actions ofmen/women cutting across various system levels Run-yan, 1988. Therefore, history is fundamentally a studyabout people, unpredictable beings who make this dis-cipline an investigation of angles and curves, rather thanlinear developments. Time and place are two crucialfactors that any historian has to consider when tryingto make sense of the experiences of the past Buttigieg,2011.

Even from these very broad definitions, the relation-ship between the two disciplines could be easily drawn.The disciplines of social psychology and history share afundamental concern with the human condition, be it inthe form of ‘individual and collective beliefs, mentalities,human behaviour and motivation, memory, personalit-ies, emotions and feelings.’ Tileaga and Byford, 2014. Inspite of the long history of mutual suspicion and inter-disciplinary uneasiness, recent works and projects areseeking to underline that the similarities between thesetwo sciences outnumber their differences, and that sub-sequently, there are more benefits to derive from dia-logue than from competition. This is the core remit ofCOST Action IS1205.

As economic historian Koji Yamamoto and social psy-

Correspondence to: S. Azzopardi (simone.azzopardi@um.edu.mt)

c© 2015 Xjenza Online

37 Azzopardi, S., Buttigieg, E., and Licata, L. (2015).Xjenza Online, 3:36–40.

chologist Vlad Glaveanu note:

Social psychologists can benefit from engagingwith historical sources by being able to con-textualise their findings and enrich their the-oretical models. . . . On the other hand, his-torians can enhance their analysis of historicalsources by drawing upon the conceptual toolsdeveloped in social psychology [and] to ‘test’these tools and contribute to their validationand enrichment from completely different per-spectives. Glaveanu and Yamamoto, 2012

The attempt of bridging history and social psycho-logy is not particular to this Action. This exercise ofbringing the two disciplines closer together is embeddedin a rich academic body of research and analysis fromboth ends of the spectrum. Scholars in social psycho-logy, such as Munsterberg (1899), Gergen (1973), Run-yan (1988), and historians such as Bloch (1924), Scott(2012), Millstone (2012) have long underlined the needof mutual exploration of the two disciplines, primarilyin the analysis of collective memory.

These previous attempts notwithstanding, the dis-tance between the two sciences remained, in that theeffort was largely made by one side only, and not sim-ultaneously. The process therefore has been fraughtwith theoretical and epistemological tensions, misunder-standings and mutual suspicion. Indeed, this is the gapthat Action IS1205 tries to bridge. By providing a solidframework and forum for experts from both fields, thisCOST Action is ensuring two-way coordination of thebridging process. Research is still underway. Yet, con-structive and promising results already point towards ahopefully successful exercise.

1.2 What is COST and Action IS1205?

COST is an intergovernmental framework for EuropeanCooperation in Science and Technology.1 Its mul-tidisciplinary element is clearly reflected in the nine keydomains which it promotes, ranging from biomedicineand molecular biosciences to transport and urban devel-opment, from physics and nano-sciences to individuals,societies, cultures and health. Action IS12052 is spe-cifically on social psychological dynamics of historicalrepresentations in the enlarged European Union and itaims at advancing knowledge on the role played by so-cial representations of history in processes of ethnic andnational constructions of European identities, and in-tergroup conflicts. The Action is stimulating scientificcooperation among social psychologists and historiansfrom around 27 European countries and Israel, along

1http://www.cost.eu/ [Last accessed: 28 Aug. 2014]2http://costis1205.wix.com/home [Last accessed: 28 Aug.

2014]

with the Palestinian Authority, New Zealand, and Ar-gentina.

The Action also plans to influence and shape both sci-entific and public domains through academic and pub-lic dissemination. Throughout the estimated four yearsallocated for this Action (2012-6), the participating his-torians and social psychologists seek to: complementand expand existing knowledge regarding the psycholo-gical processes involved in the development and main-tenance of lay representations of history; understandhow representations of the past are collectively elabor-ated and remembered; complement and expand existingsocial psychological theories of intergroup relations andprejudice by integrating a historical dimension; identify,through concerted data collection, representations of thepast that inform contemporary political conflicts acrossEurope; provide insight into how group-based emotionsinfluence processes of collective remembering, identit-ies, and intergroup relations; provide guidelines for theteaching of history of intergroup conflicts through re-viewed pedagogical methods in the teaching of history;inform political decision-makers about the influence oflay representations of history, and history education onidentities and intergroup relations through a brief policybriefing; prepare a concerted research project that willbe submitted to a series of research-funding sources,such as Horizon 2020 and Eurocores.

2 Working Groups

2.1 The Working Groups (WGs): Workin Progress

The four working groups (WGs) through which theseobjectives will be achieved share the ultimate goal of ex-ploring the interplay between lay representations of his-tory, social identities and intergroup conflict. Althoughworking relatively independently to ensure free choice ofthe most appropriate methods according to the respect-ive group objectives, the four teams are bound by a setof common tasks. All four groups set off by reviewingthe relevant literature from both disciplines in a concer-ted effort to increase mutual awareness of what alreadyexists in the respective fields. This proved crucial tothe build-up of ideas and the planning of future pro-jects within the respective remits, while also establish-ing opportunities of cooperation with the other researchgroups.

By pooling in the individual research agenda andexpertise, the participants collaborated to consolidatecommon research projects(s) which run in line with theindividual area of interests so that contribution wouldbe more forthcoming. It is noteworthy that in all WGs,more than one research project has been identified,which has motivated the formation of subgroups. These

10.7423/XJENZA.2015.1.05 www.xjenza.org

Azzopardi, S., Buttigieg, E., and Licata, L. (2015).Xjenza Online, 3:36–40. 38

have already developed or are still in the process of de-veloping research methods and designs that would con-tribute to updating the broader research agenda of eachof the WGs.

Indeed, it is an important requisite that all subgroupsof each of the WGs meet at the same time and in thesame location so that, while getting enough space towork on their own, the subgroups would be able tomutualise their research findings and build a compre-hensive and compact picture of their results. In otherinstances, two WGs may even plan to hold their meet-ings contemporaneously or to hold joint sessions wherenecessary. This in turn facilitates coordination of stud-ies among the four WGs. Such orchestrated research isthen aimed at informing the public, academics, journal-ists and decision-makers about the social psychologicalcorrelations of lay representation of history.3

2.1.1 Working Group 1

The first WG focuses on the role of social cognitive pro-cesses in shaping lay representation of history. It is co-led by social psychologists Olivier Klein, Karen Douglasand Susanne Bruckmuller and includes six subgroups.It investigates the psychological processes that explainhow communities forge collective memories around his-torical events. This in turn is meant to assist historiansin avoiding biases in historical accounts.

Subgroup 1 is concerned with ‘examining how ordin-ary people interpret historical events in a general senseby investigating what in history is considered import-ant and worthy of an explanation – and what is not.’A study has already been undertaken to gauge how ‘or-dinary’ people assess a given set of important events inEuropean history. The second part of the project fo-cuses on people’s construals of the beginning versus theending of an historical event. The initial hypothesis isthat people would attach more importance to the begin-nings of the events rather than the endings.

Subgroup 2 examines the influence of labelling a pieceof information as a ‘conspiracy theory’ on the way thatit is received. In the upcoming studies, the group plansto directly manipulate rationality by instructing par-ticipants to behave in a rational versus an intuitiveway. This is meant to further explore how ‘people’sself-concepts of rationality interact with the way inform-ation is presented’, in turn offering a deeper insight intopeople’s interpretation of political and historical events.

Subproject 3 is concerned with the epistemic and af-filiative functions of collective memory or what has beentermed ‘mnemic neglect’. The main hypothesis is thatpositive behaviours of in-group members will be bet-ter remembered than negative behaviours, especially by

3http://costis1205.wix.com/home#!working-groups/ciop[Last accessed: 28 Aug. 2014]

highly identified participants. The tool used in thisstudy is a computer-administered personality inventory,the Michigan Omnibus Personality Inventory (MOPI).Each participant answers a number of items and readsa set of statements about citizens from two differentcountries: the participant’s country and another coun-try. This survey would then explore the way peopleprocess and remember information about their own andother cultural groups.

As for subgroup 4, it is organised in two groups.The first one is concerned with the effects of exposureto counter-stereotypical narratives on attribution. Thesecond deals with the effects of exposure to counter-stereotypical narratives on cognitive processing by ex-ploring the relationship between the commitment to thegroup narrative and the level of interest in the facts de-picted by the narrative.

Subproject 5 is focused on the influence of historicalanalogies on current political judgement and attitudes.An online study on the current Ukraine conflict (Crimeacrisis) has been undertaken. The conclusions allowed fortoo many interpretations. This has given rise to severalfollow-up studies during the autumn of 2014.

As for Subproject 6, it has prepared a theoretical pa-per on historical culture.

2.1.2 Working Group 2

Social psychologist Denis Hilton, and historians ChantalKesteloot and Alberto Sa head WG2 which deals withlay representation of history in Europe, and focuses onconcepts of nationhood and identities. It is concernedwith ‘the content, structure, and properties of social rep-resentations of history, and how they relate with ethnic,national and European identities.’ A study around thisissue has already been undertaken by analysing the con-ceptions of world history from data collected from 30counties in Europe, Asia, Australasia, North and SouthAmerica Liu et al., 2012. This WG2 now aims at con-ducting a similar study on Europe only.4

Three subgroups have been formed to facilitate thisexercise, each focusing more specifically on WWI, colo-nialism and social representations of European history.Jointly, these subgroups are interested in uncovering thecommonalities and differences in representations of his-torical events across Europe; analysing how the iden-tification with the nation and with the continent cor-relates with differences in conceptions of national andEuropean history, respectively; exploring the ‘moral les-sons’ that people draw from historical events; evaluatinghow conceptions of nationhood and lay representationsof national and continental history relate, or otherwise,with existing intergroup attitudes; and exploring how

4http://costis1205.wix.com/home#!working-groups/ciop[Last accessed: 28 Aug. 2014]

10.7423/XJENZA.2015.1.05 www.xjenza.org

39 Azzopardi, S., Buttigieg, E., and Licata, L. (2015).Xjenza Online, 3:36–40.

these conceptions relate with attitudes towards immig-rants and acculturation processes in the receiving coun-tries.

The first subgroup honed in on WWI by launchinga Europe-wide survey among the participants and anyother interested third parties. The participants collec-ted/are collecting survey data about the way this waris represented nowadays, and how it relates with cur-rent attitudes and ideological positioning. This surveywill ultimately help members to draw European com-parisons, and to assess how commemorations will affectthose representations and attitudes. The data is col-lected from undergraduate psychology (n ≥ 100) andhistory (n ≥ 100) students in each participating coun-try.

Three countries (Serbia, Belgium and Finland) havecollected big enough samples to start some analysiswork. Although incomplete, preliminary conclusions areindicative of certain results. It turns out that the re-spondents have very minimal knowledge of WWI, thatthey believe that the war was the result of animosit-ies between the leaders of different nations rather thana conflict between peoples, and that a nation’s peopleand soldiers were constrained – rather than willing – tolive the war and to fight for their country.

Subgroup 2 runs on the same lines as the World His-tory Survey, but is focused on Europe instead. It is incourse of drafting a similar survey as that of the WorldHistory Survey for circulation among its participants.

The third subgroup is concerned about social repres-entations of colonialism and their implications in con-temporary intercultural relationships. Historically, co-lonialism had a profound impact on the way formerlycolonised peoples, as well as formerly colonising ones,represent themselves, the others, and the world Volpatoand Licata, 2010. However, the history of colonialismis hardly taught in most European countries. This sub-group is investigating how collective memories of thecolonial period in different settings relate with currentintergroup representations and attitudes. A text-bookanalysis in a selection of former colonised and colonisingcountries is currently underway.

2.1.3 Working Group 3

Co-led by Tibor Polya and Eva Fulop, WG3 deals withsocial-psychological studies of the narrative transmis-sion of history. The participants are chiefly working onnarratives as presented in history textbooks to studyhow institutional presentation of the past is diffusedand consumed by younger generations. The initial as-sumptions are two: history textbooks currently used inschools serve as materials of social representations; his-torical texts, because of their inevitable narrative style,are conducive to evoke empathy. Based on these twopoints, history textbooks become sources of identifica-

tion, which may project nationalism and therefore createdistortions of social identity and representation. To thiseffect, the group has selected particular events, namelyWWI and WWII and the colonial past, and a number ofselective media about them, namely novels, movies andtextbooks, to study their content and narratives.5

This WG is also using computer-assisted techniques oftext analysis. This is the linguistic development envir-onment NOOJ that includes large-coverage dictionaries,grammars and parses corpora in real time. Dictionariesand grammars are applied to texts in order to locatemorphological, lexical and syntactic patterns and tagsimple and compound words. NOOJ dictionaries andgrammars can be built by users and they can processa dozen languages, including some Roman, Germanic,Slavic, Semitic and Asian languages, as well as Hun-garian. It will be used to analyse various psychologicaldimensions in history textbooks.6

2.1.4 Working Group 4

The fourth and final WG, headed by Micha l Bilewicz& Sabina Cehajic-Clancy, covers the roles of lay repres-entations of history and group-based emotions in inter-group conflict and reconciliation processes. This groupis working on the interplay between such specific history-related emotions as collective guilt, shame and pride,and intergroup mediation or confrontation.

Research on collective guilt has already yielded im-portant results. What WG4 is in the course of doingis to extend this research further to cover the follow-ing areas: the conditions under which intergroup apo-logies and reparations satisfy their target groups; whysome people defend against national guilt and othersaccept it; how victimisation influences intergroup atti-tudes; and how to temper the role of historical moralschemes in present day political attitudes.7

WG4 has established several project lines: to explorethe extent to which and how intergroup relations inethnically-mixed countries are determined by the dif-ferent historical beliefs about ownership or authoctony;to look into the adherence to the ‘official’ historical nar-rative; to evaluate the dimensions of national identific-ation; to assess the impact of official apologies on vic-tim or perpetrator group members; or to investigate therole played by historical moral exemplars in historicalnarratives. Besides this, WG4 has also performed sev-eral studies in countries where historical genocides tookplace (Bosnia & Herzegovina, Poland), looking into howperpetrator, victim and bystander groups construe thepast, with a view to analyse which strategies can be em-

5http://costis1205.wix.com/home#!working-groups/ciop[Last accessed: 28 Aug. 2014]

6www.noojnlp.net/ [Last accessed: 28 Aug. 2014]7http://costis1205.wix.com/home#!working-groups/ciop

[Last accessed: 28 Aug. 2014]

10.7423/XJENZA.2015.1.05 www.xjenza.org

Azzopardi, S., Buttigieg, E., and Licata, L. (2015).Xjenza Online, 3:36–40. 40

ployed to foster reconciliation between them. The mainsubgroups deal with the following specified areas: vic-tims and perpetrators - contemporary social perceptionsof Fascism in different European countries; responsibil-ity displacement; moral exemplars in history; historicalrepresentations of the Holocaust; parrhesiastic reconcili-ation.

3 ConclusionThe interaction of history and social psychology in thisAction is seeking to find ways in which the past can bebetter understood in the present, while acknowledgingthat the present is forever informing perceptions of thepast. The work in progress is successfully exploring thehealthy symbiosis between the two fields and the parti-cipants are already reaping the benefits of this intens-ive dialogue. Some limitations have emerged, such asthe numerical imbalance of social psychologists over his-torians, and the theoretical versus the interpretationalpriorities of social psychologists and historians, respect-ively. However, the increasing mutual awareness of thecommon grounds between the two disciplines means thatthe participants are constantly seeking to work on thestrong points between them in search of the broader ob-jective of Action IS1205.

AcknowledgementsThis work was conducted within the framework ofCOST Action IS1205: Social psychological dynamicsof historical representations in the enlarged EuropeanUnion.

The authors would like to thank the working groupleaders of Action IS1205 for conceding permission to usethe unpublished meeting reports and minutes.

References

Bloch, M. (1924). Les Rois Thaumaturges: etude sur leCaractere Surnaturel Attribue a la Puissance Roy-ale, Particulierement en France et en Angleterre(1st ed.). Strasbourg: Librairie Istra.

Buttigieg, E. (2011). Nobility, Faith and Masculinity.The Hospitaller Knights of Malta, c.1580-c.1700(1st ed.). London: Continuum.

Gergen, K. J. (1973). Social psychology as history. J.Pers. Soc. Psychol. 26 (2), 309–320.

Glaveanu, V. & Yamamoto, K. (2012). Bridging Historyand Social Psychology: What, How and Why. In-tegr. Psychol. Behav. Sci. 46 (4), 431–439.

Liu, J. H., Paez, D., Hanke, K., Rosa, A., Hilton,D. J., Sibley, C. G., . . . Suwa, K.-i. (2012). Cross-Cultural Dimensions of Meaning in the Evaluationof Events in World History?: Perceptions of His-torical Calamities and Progress in Cross-CulturalData From Thirty Societies. J. Cross. Cult. Psy-chol. 43 (2), 251–272.

Millstone, N. (2012). ’Historicising Common Sense’. In-tegr. Psychol. Behav. Sci. 46 (4), 529–543.

Munsterberg, H. (1899). Psychology and History. Pres-ident’s Address, American Psychological Associ-ation, New York Meeting, December, 1898. Psy-chol. Rev. 6 (1), 1–33.

Runyan, W. M. (1988). Psychology and Historical Inter-pretation. Oxford: Oxford University Press.

Scott, J. W. (2012). The incommensurability of psycho-analysis and history. Hist. Theory, 51 (1), 63–83.

Tileaga, C. & Byford, J. (2014). Psychology and History:Interdisciplinary Explorations. Cambridge Univer-sity Press.

Volpato, C. & Licata, L. (2010). Collective memories ofcolonial violence: Introduction to the Focus sec-tion. Int. J. Conf. Violence, 4 (1), 4–10.

10.7423/XJENZA.2015.1.05 www.xjenza.org

Sultana, D. (2015).Xjenza Online, 3:41–50.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.06

Research Article

Numerical Modelling of Soil Erosion Susceptibility in the MalteseIslands using Geographic Information Systems and the Revised Uni-versal Soil Loss Equation (RUSLE)

D. SultanaMalta Environment and Planning Authority, St Francis Ravelin, Floriana, FRN1230, Malta

Abstract. The Mediterranean region is subject tovarious factors that exacerbate soil erosion pressures.Such factors include agricultural land fragmentation andabandonment, unsustainable agricultural practices andrapid urbanisation. Soil erosion in the Maltese Islandshas been identified as a predominating land degrada-tion process and a major threat to the sustainability ofthe agricultural sector. The small scale of the MalteseIslands facilitates an in detail national study of soilerosion processes and contributing socio-economic dy-namics. The research methods, erosion rate values andcontrolling dynamics discussed in this work have a par-ticular relevance to the Mediterranean area.

1 Introduction

Soil erosion is triggered by a combination of natural andanthropogenic factors that include steep slope gradients,intense precipitation, low vegetation cover and inappro-priate land use (Renschler, Mannaerts & Diekkruger,1999; Wischmeier & Smith, 1978). Prolonged erosionleads to an irreversible loss of ecological and agricul-tural soil function and associated ecosystems services.Erosion reduces agricultural productivity, posing limit-ations to sustainable agricultural use. The aspect of soilerosion that causes most concern is the loss of topsoil,the most fertile part of the soil profile (Gobin et al.,2004).

The Global Assessment of Human-induced Soil De-gradation (GLASOD) map estimates that 114 millionhectares are affected by human-induced soil erosion(Oldeman, Hakkeling & Sombroek, 1991). The principaldrivers of soil erosion in the European Union are unsus-tainable agricultural practices, overgrazing, deforesta-tion and construction activities (Oldeman et al., 1991).

The Mediterranean region is particularly susceptible toerosion (EEA, 1999). High erosion rates, in conjunctionto slow soil formation, lead to irreversible reductions ofMediterranean soil quality and quantity.

The Maltese Islands (Figure 1) are located in thecentre of the Mediterranean Sea. The Islands have atotal land area of 316 km2 and comprise three main is-lands, Malta, Gozo and Comino, and a number of outly-ing islets.

Figure 1: Map of the Maltese Islands (from (Ezilon, 2009))

The Maltese Islands have a semi-arid Mediterraneanclimate, with mild, wet winters and hot, dry summers.The average annual rainfall is around 524 mm and theaverage yearly temperature is 22.5 C. Rainfall is char-acterised by storms of high intensity and relatively shortduration (Government of Malta, 2002).

Correspondence to: D. Sultana (daniel.sultana@um.edu.mt)

c© 2015 Xjenza Online

Sultana, D. (2015).Xjenza Online, 3:41–50. 42

The Maltese Islands, and indeed the rest of the Medi-terranean, are subject to various local factors which ex-acerbate soil erosion pressures. Such factors include ag-ricultural land fragmentation and abandonment, limitedsoil agriculture suitability, unsustainable agriculturalpractice, rapid urbanisation, limited water resources andrapidly modernising social structure. As a consequenceof the above interacting factors, soil erosion has beenidentified as a predominating land degradation processand a major threat to the sustainability of the agricul-tural sector (Tanti, Role, Borg & Calleja, 2002).

Maltese soil erosion risk modelling predating this art-icle consisted of a numerical model developed by Tantiet al. (2002) assessing the northwestern region of Malta.The model identified areas threatened by soil erosion onthe basis of geological substrate, slope, retaining rubblewalls state and land cover. Model results clearly in-dicated that the assessed area was subject to high soilerosion rates (Tanti et al., 2002).

Presently, a large variety of empirical, semi-empirical,and physical process-based soil erosion risk models areavailable (Gitas, Douros, Minakou & Silleos, 2009; Erkal& Yildirim, 2012). The most widely applied empiricalmodel (Fistikoglu & Harmancioglu, 2002) for assessingsoil erosion by water driven mechanisms is the UniversalSoil Loss Equation (USLE), developed by Wischmeierand Smith (1978). The USLE and its revision RUSLE(Renard, Foster, Weesies, McCool & Yoder, 1997) applymore than 40 years of experimental field observationsgathered by the Agricultural Research Service of theUSDA (Novotny & Olem, 1994). The RUSLE is appliedis this study.

The dynamic relationship between human activitiesand resulting soil erosion requires that erosion be mon-itored. Regular monitoring allows competent author-ities to appreciate the influence policy and land usechange mechanisms have on soil erosion. Our studyaims to provide quantitative estimates of soil erosionby water of the Maltese Islands for the year 2013.The discussion section examines the interaction betweenthe socio-economic situation and consequent effects onsoil erosion. This approach ties environmental scienceto policy issues and provides an integrated approachthrough which professionals and government may prior-itise and present context specific erosion control meas-ures. In this framework, high erosion risk areas aresingled out, the physical, socio-economic and policymechanisms influencing the area identified, and erosioncontrol measures, via policy and physical intervention,suggested and implemented to reduce risk.

2 Methods

The RUSLE technique was applied and built into aGIS-based model. Relevant model input parameters

were prepared separately and stored as GIS vector lay-ers. Five vector layers, each representative of RUSLEfactors, were converted to raster layers with a grid res-olution of 50 metres. Each cell has a value representat-ive of the area’s factor value. Each raster layer was thencombined in the GIS model to calculate soil loss for eachcell in the study area for the year 2013. The predictedsoil losses were verified against field observations of soilerosion made at the end of the 2013 winter season. Thesection below discusses the method followed to obtainthe five RUSLE factor values.

2.1 RUSLE factors

The factors assessed in the RUSLE are rainfall erosiv-ity (R), soil erodibility (K), slope length and steepness(LS), cover and management practices (C) and conser-vation practices (P) (Wischmeier & Smith, 1978). Thesefactors are combined in a numerical formula (equation1). The computation returns soil loss per unit area,equivalent to predicted erosion in ton hectare−1 year−1

(Gitas et al., 2009).

A = R×K × LS × C × P (1)

where A = average annual soil loss (t ha−1 yr−1), R= rainfall/runoff erosivity (MJ mm ha−1 h−1yr−1), K =soil erodibility (t h MJ−1mm−1), LS = slope length andsteepness (dimensionless), C = cover management (di-mensionless), P = support practice (dimensionless)

2.1.1 Rainfall Erosivity (R factor)

Rainfall erosivity is a climatic factor that takes intoaccount the erosive capacity of rainfall (D’Odorico,Yoo & Over, 2001; Le Bissonnais, Montier, Jamagne,Daroussin & King, 2002). The factor is determinedas a function of total storm kinetic energy (E) andits maximum 30-min intensity (Imax30) (Wischmeier,1959; Wischmeier & Smith, 1958).

The authors refer to the method employed by (Ir-aldo et al., 2013) in the calculation of the Maltese Rfactor. Iraldo et al. (2013) applied a modification ofthe Fournier index F , developed by (Arnoldus, 1980),the modified F index (FF ) (Ferro, Porto & Yu, 1999)(equation 2). The method uses average monthly (pij)and annual precipitation (P). This approach is thoughtto be better correlated with rainfall erosivity.

FF =1

N

N∑j=1

12∑i=1

(p2ij

Pj

)(2)

where pij is the rainfall in month (mm) of the year j andP is the total rainfall per year.

Iraldo et al. (2013) calculate the FF index usingrainfall data for Malta (http://www.maltaweather.com)over the period 1985–2012. The R value was estimated

10.7423/XJENZA.2015.1.06 www.xjenza.org

43 Sultana, D. (2015).Xjenza Online, 3:41–50.

using the average of R: FF relationships adapted for Si-cily (Ferro et al., 1999; Renard & Freimund, 1994). R =0.612 F 1.56 (Sicily). The R factor value defined by Iraldoet al. (2013) for Malta was 832.16 MJ mm ha−1 h−1 yr−1.

2.1.2 Soil erodibility (K factor)

Soil erodibility expresses the intrinsic capacity of the soilto be eroded and reflects the effect of the average long-term soil profile response to rainfall and runoff erosion.The main soil properties affecting K are soil texture,organic matter, structure, and soil permeability (Erkal& Yildirim, 2012). High organic content decreases soilerodibility (F.A.O, 1996).

An extensive national soil survey, assessing over threehundred data points in a grid distribution of 1 km spa-cing, was carried out in 2002 (MALta Soil InformationSystem, 2003). Amongst other soil parameters, the sur-vey defined organic matter content and texture. Fol-lowing Table 1, the MALSIS (2002) data set was usedto define two hundred and sixty eight K factor values(Figure 2) for the Maltese Islands. Soil erodibility factor(K) value varies from 0.12 to 0.6 t ha h MJ−1 mm−1 andthe mean value is 0.29 t ha h MJ−1 mm−1.

Table 1: K values as they were calibrated according tospecific soil parameters

Textural classes Organicmatter lessthan 2 %

Organicmatter morethan 2 %

Clay 0.24 0.21Clay Loam 0.33 0.28Coarse Sand Loam / 0.07Fine Sand 0.09 0.06Fine Sandy Loam 0.22 0.17Heavy Clay 0.19 0.15Loam 0.34 0.26Loamy Fine Sand 0.15 0.09Loamy Sand 0.05 0.04Loamy Very Fine Sand 0.44 0.25Sand 0.03 0.01Sandy Clay Loam / 0.20Sandy Loam 0.14 0.12Silty Loam 0.41 0.37Silty Clay 0.27 0.26Silty Clay Loam 0.35 0.30Very Fine Sand 0.46 0.37Very Fine Sandy Loam 0.41 0.30Loamy sand 0.05 0.04Silt 0.43 0.60Sandy clay 0.10 0.14Clay and heavy clay 0.24 0.21

The data points were then interpolated following GIS

kriging method. The resulting GIS layer was convertedto a raster map with a grid resolution of 50 metres. Eachcell has a value representative of the area’s K factorvalue (Figure 2).

Figure 2: K factor raster map showing K factor values.

2.1.3 Slope length and steepness (LS factor)

Slope length and steepness reflect the proportional effecttopography has on erosion (Foster & Wischmeier, 1974;Wischmeier & Smith, 1978). For this study, the LSfactor was computed from a Digital Elevation Model(DEM) with the ArcGIS Spatial Analyst extension. TheLS factor was calculated at a 10 m horizontal spacingfrom a 1:1500 scale topographic DEM following equation3.

LS =

(Slope− length

22.1

)0.5

0.065

+ 0.0456(slope) + 0.00654(slope)2 (3)

where Slope−length is in meters and slope is in %.The computed LS factor GIS layer was converted to

a raster map with a grid resolution of 50 metres. Eachcell has a value representative of the area’s LS factorvalue (Figure 3).

2.1.4 Crop and Vegetation management (Cfactor)

The management factor reflects the effect cropping andmanagement practices have on erosion rates. The Cfactor is closely linked to land-use types and is a factorin soil erosion vulnerability (Wischmeier & Smith, 1978;Beskow et al., 2009). For this study, the C factor of thestudy area was obtained from a high resolution aerialorthophoto set produced in June 2012. The orthophotodata set was manually interpreted by the author, a landcover expert (Figure 4). Ground truthing surveys and

10.7423/XJENZA.2015.1.06 www.xjenza.org

Sultana, D. (2015).Xjenza Online, 3:41–50. 44

Figure 3: LS factor raster map.

land cover reports were consulted as a means of ensuringcorrect orthophoto land cover interpretation. A min-imum mapping unit of 10.000 m2 was applied followinga classification system compliant to the CLC 2006 tech-nical guidelines. A number of additional layers weredefined in view of the large scale of orthophoto inter-pretation (1:5000). Maltese Island cover was dividedinto twenty two land-use types (Figure 4).

Figure 4: Detailed map showing spatial distribution of landuse and cover in the Maltese Islands.

C factor values, corresponding to land cover classes(Table 2), were defined following expert defined de-scriptions (e.g., Wischmeier and Smith (1978), Mor-gan (1995), Pasak, Janecek and Sabata (1983), Alena(1991), Malısek (1992). The classification scheme ex-cludes urbanised areas, bare rocks and water surfacesfrom evaluation since these surfaces contain no soil.Discontinuous urban, green urban (semi-permeable sur-faces), mixed forest (moderate soil cover), maquis (with

little to no soil cover), beaches, dunes, sand plains, rockysteppe and salt marshes as cover types of good manage-ment practice with a C factor value of less than 0.1.Maquis (with moderate soil cover), green urban (semi-permeable surfaces), sports and leisure (semi-permeablesurfaces), pastures and permanent crops are categorisedas land cover of moderate management practice with adesignated C factor value between 0.11 and 0.4. Ar-able land, abandoned and degraded agricultural areas,and degraded semi-natural areas are categorised as landcover or low management practice of C factor valuesbetween 0.41 and 0.8. Land-use classes were allocatedC values without considering seasonal variance.

Agricultural practice is of particular relevance whendefining agricultural C factor. A large portion ofMaltese arable soils are exposed, have no vegetationcover, and are deep-ploughed in anticipation of the firsttorrential September rains (RDP, 2007–2013). This ag-ricultural practice intensifies water erosion and, as a con-sequence, a high C factor value was assigned to areascovered by arable land. Abandoned/ degraded agricul-tural areas and degraded semi-natural areas have thehighest K factor values. Vegetation cover in these areasis often entirely removed to accommodate bird trap-ping, parking and other such activities that contributetowards accelerated soil erosion.

The land cover dependent C factor values weremapped in GIS. The resulting layer was converted to araster map with a grid resolution of 50 metres. Each cellhas a value representative of the area’s C factor value(Figure 5).

Figure 5: C factor raster map.

2.1.5 Erosion control (P factor)

Erosion control represents the effects various prac-tices have on preventing soil erosion by water run-off. Wischmeier and Smith (1978) discuss that con-trol structures, which include inter alia improved tillage

10.7423/XJENZA.2015.1.06 www.xjenza.org

45 Sultana, D. (2015).Xjenza Online, 3:41–50.

Table 2: Land-use classes, cover in km2 and allocated C value (land-use listed below only includes those with a C value)

Land cover type Cover km2 (%of total land)

C factor value

Discontinuous urban 3.36 (1.06%) 0.10Green urban (semi-permeable surfaces) 0.26 (0.08%) 0.15Sports + Leisure (semi-permeable surfaces) 1.95 (0.62%) 0.20Arable land 164.50 (52.1%) 0.50Permanent crops 2.91 (0.92%) 0.40Abandoned + Degraded Agricultural areas 9.56 (3.03%) 0.80Mixed forest (moderate soil cover) 4.20 (1.33%) 0.10Steppe + Garrigue 36.07 (11.42%) 0.25Maquis (with moderate soil cover) 4.69 (1.49%) 0.20Maquis (with little to no soil cover) 0.08 (0.03%) 0.15Beaches, dunes, sand plains 0.12 (0.04%) 0.10Bare rock / rocky steppe (little to no soil cover) 5.83 (1.83%) 0.05Degraded semi-natural areas 7.28 (2.31%) 0.60Salt marshes 0.15 (0.05%) 0.05

practices, strip cropping and terraces, should signific-antly contribute towards erosion control and frequentlyprovide the major control in a farmer’s field. The lowerthe P values, the more effective the conservation prac-tice is deemed to be in reducing soil erosion (Erkal &Yildirim, 2012).

Appropriate farming practices may positively influ-ence countryside and landscape quality, and sustainkey environmental resources such as biodiversity, soiland water. Terraced agricultural fields are recognisedas a characteristic feature of Mediterranean landscapes(Whitelaw & French, 1999; Frederick & Krahtopoulou,2000; Grove & Rackham, 2001; Price & Nixon, 2005).Terraces adjust hillslopes into stepped, contour parallel,agricultural units of relatively flat ground suitable forcultivation. In the Mediterranean, terrace constructionhas typically involved the use of interlocked dry stonerisers, rubble walls. These walls act as retainers to sup-port back-lying beds of level soil. Although the originalpurpose of terrace construction is the increase of agricul-tural areas, these structures provide a necessary meansof soil erosion control (Bevan & Conolly, 2011).

Maltese agricultural practices have significant controlon agricultural land susceptibility to degradation andsoil erosion. Tanti et al. (2002) identify retaining rubblewalls in terraced fields as the most important water andsoil erosion control method structures in the MalteseIslands (Tanti et al., 2002). Contour ploughing was alsoidentified as a key erosion mitigation practice.

A national survey was conducted by the author overfour months, starting in June 2013. The survey assessedrubble wall state following a classification scheme con-sisting of three potential rubble wall states. In this clas-sification scheme, rubble walls in a good state contain

a maximum of 1 breach showing half of the soil pro-file; rubble walls in moderate state contain more than 1breach, but no more than 3, showing half of the soil pro-file, and rubble walls in a poor state contain more than3 breaches showing half of the soil profile or 1 or morethat show the whole soil profile. Each state is attributedP factor values; good state P factor value 0.3, moder-ate state P factor value 0.5, and poor state P factorvalue 0.7. The resulting GIS rubble wall state layer wasconverted to a raster map with a grid resolution of 50metres. Each cell has a value representative of the area’sP factor value (Figure 6).

Figure 6: P factor raster map. Blank (white) areas repres-ent urban areas or areas that contain no rubble walls.

10.7423/XJENZA.2015.1.06 www.xjenza.org

Sultana, D. (2015).Xjenza Online, 3:41–50. 46

3 Results

3.1 Computed soil loss

The average annual soil loss was computed on a cell-by-cell basis following equation 1. The five factor ras-ter maps representing R,K,LS,C and P factors, wereoverlain and multiplied with the ArcGIS Spatial Ana-lyst extension. The erosion map (Figure 7) shows thespatial distribution of soil loss in the Maltese Islands ex-pressed as annual average soil loss in tonnes per hectareper year. The values should however be considered in acomparative manner rather than absolute values. Thisis due to the generalisation of the used input data aswell as the nature of the model.

In order to obtain a better general understanding andbe able to carry out a National comparison, the quantit-ative output of soil loss prediction was classified in eightcategories of increasing soil loss severity: < 1 (none), 1to 2, 2 to 5 (very low), 5 to 10 (low), 10 to 25 (moderate),25 to 45 (high), 45 to 75 (very high), > 75 t ha−1 yr−1

(severe). Erosion severity thresholds are consistent withthose presented by various experts (e.g., Iraldo et al.(2013)). Such a classification is consistent with theRUSLE model’s role as a conservation management tool,where relative comparisons among areas are more signi-ficant than any assessment of the absolute soil loss in aparticular location.

Figure 7: Average annual soil loss (t ha−1 yr−1) in theMaltese Islands following RUSLE equation.

3.2 Areas at risk

Calculated National annual soil loss (Figure 7 and 8)indicates that 61.01 km2, 19.33 % of total National landarea, are at risk of moderate (10 to 25 t ha−1 yr−1) tosevere (> 75 t ha−1 yr−1) soil erosion.

Figure 8: Soil loss potential histogram. X-axis numberof cells (2500 m2) showing erosion values that fall withinerosion rate categories, Y-axis soil erosion rate categories.

4 Discussion

Maltese central and north-eastern areas show the low-est erosion risk. These areas are characterised by re-latively flat topographies, good land management anderosion control measures. Maltese north-western andGozitan areas are characterised by a large range inerosion rates. Within this area, low erosion risk occursin plateaus comprising low topographic gradients, andthe application of good land management and erosioncontrol measures. Plateau flanks typically consist ofexceptionally high erosion rates, characterised by hightopographic gradients, inappropriate cultivation prac-tices and poor erosion control measures. Steeply in-clined plateau flanks demonstrating low erosion risk areassociated with areas demonstrating adequate vegeta-tion cover, and effective management and conservationpractices.

The spatial pattern of modelled potential soil erosion(Figure 7) is clearly proportional to slope gradient. Fieldwork and national reports also identify land use (Cfactor) and control measures (P factor) as being critical.The highest estimates of quantitatively measured andpredicted erosion rates occur in steeply inclined arableland where poor management and conservation prac-tices are applied (e.g., Tanti et al. (2002).

A large portion of Maltese arable soils are exposed, allvegetation cover is removed and deep-ploughed in anti-cipation of the first torrential September rains. This ag-ricultural practice intensifies water erosion. The authorproposes that strip contour ploughing, where vegetationcover is retained between ploughed areas, is applied toreduce water induced soil erosion. Field evidence alsoclearly identifies agriculture retaining rubble walls as akey soil erosion control method in steeply inclined ag-riculture areas. The reasons leading to inappropriateagricultural practices, and consequent soil erosion, areclosely tied to the National socio-economic situation of-ten common to the Mediterranean region.

Agricultural land ownership is a key issue in the

10.7423/XJENZA.2015.1.06 www.xjenza.org

47 Sultana, D. (2015).Xjenza Online, 3:41–50.

Maltese Islands. Two thirds of the agricultural landis owned by the State and the remaining one-third bythe private sector. Eighty percent of cultivated agri-cultural land is rented and twenty percent is occupiedby owners or under a freehold basis (National Statist-ics Office, 2003). In accordance with the AgriculturalLeases (Re-letting) Act drawn up in 1967, governmentand privately owned land is automatically re-let to theexisting tenant or descendants. Law impedes the evic-tion of tenants or any substantial increases of rent, evenon privately owned land. Given the low prices at whichland is rented and prospects of strong land speculation,both tenants and private landowners tend to hold on toleased land. This situation has significant national con-sequences. Should leased agricultural land not providea significant source of income, the arable land may bedisused or used for other purposes. This process leads toaccelerated land degradation and reduces the economiesof scale potential, reduced production, of the Maltese ag-ricultural sector (Rural Development Plan, 2007–2013).

Statistical results from the 2010 census indicate thatseventy four percent (9.203 ha) of all agricultural hold-ings cover less than one hectare (census, 2010). Signi-ficantly contributing to this issue is land fragmentation,brought about by inheritance and parcel sale. This pro-cess significantly reduces the total exploitable land andthus diminishes economic viability of agriculture pro-duction. This may consequentially lead to land aban-donment or change in land use.

Land abandonment may also be the consequence ofincreased international agricultural product cost com-petitiveness. Mediterranean agriculture is faced withsevere limitations in this area. Naveh (1991) estimatesthat more than half of Mediterranean land is of mar-ginal economic agricultural potential, characterised bysteep, rocky uplands and poor soils. This setting of-ten presents insurmountable economic obstacles for theintroduction of modern agricultural techniques neces-sary in modern markets (Pinto Correia, 1993). For theMediterranean regions, the trend towards land abandon-ment is accentuated by increasing competition with thehighly productive agriculture of northwestern Europe(Pinto Correia, 1993). Agricultural activity survived inthe Maltese Islands in the past fifty years as a result ofprotective measures, namely price guarantees and quotarestrictions on imports, aimed at encouraging produc-tion by ensuring a regular income flow for local farmers.Maltese entry to the EU (2004) led to the dismantlingof various protective levies and extensive sector restruc-turing to adhere to EU legislation (RDP, 2007–2013).These developments adversely influenced net farmer in-come. As a consequence, landowners may have to sus-tain net income through alternative employment, lead-ing to reduced land management and land degradation.

When agricultural exploits do not produce a source ofrevenue, farmers may also resign and abandon the areaaltogether.

The socio-economic conditions discussed above, ag-ricultural land ownership, increased international agri-cultural product cost, agricultural holding size and landfragmentation, constrain net farm income. The gener-ation of income via agricultural practices is central tounderstanding whether agricultural land is used and in-vested upon or abandoned and erosion processes poten-tially intensified.

The effects of land abandonment on soil quality andsoil erosion may be either positive or negative. The keycontrol on soil regenerative capacity is vegetation cover,controlled by climatic conditions and soil quality. Nu-merous authors have demonstrated that in a wide rangeof environments both runoff and sediment loss decreaseexponentially as the percentage of vegetation coverincreases (e.g., Elwell and Stocking (1976), Lee andSkogerboe (1985), Francis and Thornes (1990). Con-sequently, should the applied agricultural managementpractices have been unfavourable, the re-establishmentof natural vegetation cover may reduce soil erosion.Central Mediterranean climatic conditions, dry sum-mers reduce vegetation cover and winter flash floodsexacerbate soil erosion, however do not favour naturalvegetation reclamation. Another key parameter, signi-ficantly contributing to soil erosion in terraced fields, isthe degradation of soil retaining rubble walls.

Soil retaining rubble walls in terraced fields are an-thropogenic structures characteristic of sloped Mediter-ranean agricultural areas. Although the original pur-pose of terrace construction is the increase of agricul-tural areas, these structures provide a necessary meansof soil erosion control (Bevan & Conolly, 2011). InMalta, significant expanses of garrigue were reclaimedfor agricultural use; rubble material was used for level-ling, topped-off with soil and retained by rubble walls(Rural Development Plan, 2004-2007). Under naturalconditions, these soils would not accumulate in suchareas, and given the opportunity, gravitational processeswould transport soils to more stable areas. Unfavour-able climatic conditions hinder natural vegetation re-clamation and as a consequence, once rubble walls arebreached and not restored, intensive soil erosion occurs.

The Maltese National survey, carried out by the au-thor assessing land use (Figure 4) and rubble wall state(Figure 6), concludes that the majority of agriculturalterraces on inclined surfaces are disused and retainingrubble walls in a derelict state. These steeply inclinedagricultural fields show the highest National soil erosionrates (Figure 7). These areas are subject to varioussocio-economic conditions that constrain net farm in-come. Such hindering conditions include agricultural

10.7423/XJENZA.2015.1.06 www.xjenza.org

Sultana, D. (2015).Xjenza Online, 3:41–50. 48

land ownership legislation, increased international agri-cultural competition, agricultural holding size and landfragmentation. In the Maltese Islands, these fields, oflimited size, low accessibility, and requiring high rubblewall maintenance, may have once been economically ex-ploitable. However, with changes in socio-economic dy-namics, the economic incentive for tending these mar-ginal fields was lost and the fields abandoned. It is pro-posed that in most cases, these socio-economic factors,common to Mediterranean countries, significantly con-tributed towards agricultural land disuse and dilapida-tion of soil retaining structures. These conditions arethe main drivers leading to accelerated soil loss. Thesocio-economic dynamics, and the consequent effectson agricultural practices and soil erosion processes dis-cussed in this study are characteristic of the Mediter-ranean region.

There is an urgent need for an update of national le-gislation to alleviate the adverse effects socio-economicparameters have on agricultural practices. The authorproposes that ameliorating net farmer income will dir-ectly increase agricultural land use, reversing the currentabandonment trend. This mechanism will indirectly in-crease the maintenance of key soil erosion control struc-tures, the terraced field rubble wall, and reduce the cur-rent alarming rate of soil erosion.

5 Conclusion

Calculated National annual soil loss (Figure 7) indic-ates that 61.01 km2, 19.33 % of total National land area,are at risk of moderate to severe soil erosion. Maltesecentral and north-eastern areas show the lowest erosionrisk. These areas are characterised by relatively flat to-pographies, good land management and erosion controlmeasures. Maltese north-western and Gozitan areas arecharacterised by a large range in erosion rates. Withinthis area, low erosion risk occurs in plateaus comprisinglow topographic gradients, and the application of goodland management and erosion control measures. Plat-eau flanks typically consist of exceptionally high erosionrates, characterised by high topographic gradients, inap-propriate cultivation practices and poor erosion controlmeasures. Steeply inclined plateau flanks demonstrat-ing low erosion risk are associated with areas demon-strating adequate vegetation cover, and effective man-agement and conservation practices.

Analysis of the Maltese National land cover (Figure 4)and rubble wall state (Figure 6) survey, carried out bythe authors, concludes that the majority of agriculturalterraces on inclined surfaces are in a derelict state andin most cases disused. These steeply inclined agricul-tural fields show the highest National soil erosion rates(Figure 7). These areas are subject to various socio-economic conditions that constrain net farm income. We

propose that in most cases socio-economic factors, com-mon to Mediterranean countries, significantly contrib-ute towards agricultural land disuse, dilapidation of soilretaining structures and accelerated soil erosion.

Soil is a limited resource in the Mediterranean areaboth in terms of quantity and quality. Soil resourcessupport agriculture, maintain ecosystem health and arecentral to hydrological processes. Although of great im-portance, soil resources are relatively mismanaged andare threatened by accelerated erosion rates. There is anurgent need for an update of national legislation to alle-viate the adverse effects socio-economic parameters haveon agricultural practices. Ameliorating Mediterraneannet farmer income will directly increase agricultural landuse, reversing the current abandonment trend. Thismechanism will indirectly increase the maintenance ofkey soil erosion control structures, the terraced fieldrubble wall, and reduce the current alarming rate of soilerosion.

AcknowledgementsI wish to thank the European Cooperation in Sci-ence and Technology (COST), Stephen Conchin forhis significant contribution towards GIS analysis, Dar-rin Stevens from the Malta Environment and Plan-ning Authority (MEPA) for a number of fruitful discus-sions, Emmy Donkers from the University of Wagenin-gen (Netherlands), Josh Copping and John Kenworthyfrom the Manchester Metropolitan University (England)and Francesca Scerri, Gabriel Farrugia and Daniel Vellafrom the University of Malta for their help in data gath-ering and processing.

References

Alena, F. (1991). Protierozna ochrana na ornej pode(Protection of arable soils against erosion). Con-ceptual Study. Bratislava: Statna melioracnasprava.

Arnoldus, H. M. J. (1980). An approximation of therainfall factor in the Universal Soil Loss Equation.In M. de Boodt & D. Gabriels (Eds.), Assess. eros.proc. work. assess. eros. usa eur. (pp. 127–132).Chichester: Wiley & Sons.

Beskow, S., Mello, C. R., Norton, L. D., Curi, N., Viola,M. R. & Avanzi, J. C. (2009). Soil erosion pre-diction in the Grande River Basin, Brazil usingdistributed modeling. CATENA, 79 (1), 49–59.

Bevan, A. & Conolly, J. (2011). Terraced fields andMediterranean landscape structure: An analyt-ical case study from Antikythera, Greece. J. Ecol.Model. 222 (7), 1303–1314.

census, M. (2010). Rural Development Plan for Malta,2004-2006. (2004). Rural Development Depart-

10.7423/XJENZA.2015.1.06 www.xjenza.org

49 Sultana, D. (2015).Xjenza Online, 3:41–50.

ment Ministry for Rural Affairs and the Environ-ment.

D’Odorico, P., Yoo, J. & Over, T. M. (2001). An assess-ment of ENSO-induced patterns of rainfall erosiv-ity in the Southwestern United States. J. Clim.14 (21), 4230–4242.

EEA. (1999). Environment in the European Union atthe turn of the century. European EnvironmentAgency. Copenhagen.

Elwell, H. A. & Stocking, M. A. (1976). Vegetal coverto estimate soil erosion hazard in Rhodesia. Geo-derma, 15 (1), 61–70.

Erkal, T. & Yildirim, U. (2012). Soil Erosion Risk As-sessment in the Sincanli Sub-Watershed of theAkarcay Basin (Afyonkarahisar, Turkey) Using theUniversal Soil Loss Equation (USLE). Ekoloji,21 (84), 18–29.

Ezilon. (2009). The physical map of malta showing ma-jor geographical features like elevations, islands,mountain ranges, seas, lakes, plateaus, peninsulas,rivers, plains, landforms and other topographicfeatures. Retrieved July 10, 2015, from %5Curl%7Bhttp://www.ezilon.com/maps/europe/malta-physical-maps.html%7D

F.A.O. (1996). A Provisional Methodology For Soil De-gradation Assessment. Rome.

Ferro, V., Porto, P. & Yu, B. (1999). A comparativestudy of rainfall erosivity estimation for southernItaly and southeastern Australia. Hydrol. Sci. J.44 (1), 3–24.

Fistikoglu, O. & Harmancioglu, N. B. (2002). Integ-ration of GIS with USLE in Assessment of SoilErosion. Water Resour. Manag. 16 (6), 447–467.

Foster, G. R. & Wischmeier, W. H. (1974). Evaluatingirregular slopes for soil loss prediction. Trans. Am.Soc. Agric. Eng. 17 (2), 305–309.

Francis, C. F. & Thornes, J. B. (1990). Runoff hydro-graphs from three Mediterranean vegetation covertypes. In J. B. Thornes (Ed.), Veg. erosion, pro-cess. environ. (pp. 363–384). Chichester: Wiley &Sons.

Frederick, C. & Krahtopoulou, A. (2000). Deconstruct-ing agricultural terraces; examining the influenceof construction method on stratigraphy, datingand archaeological visibility. In P. Halstead & C.Frederick (Eds.), Landsc. l. use postglacial greece(pp. 79–94). Sheffield: Sheffield Academic Press.

Gitas, I. Z., Douros, K., Minakou, C. & Silleos, G. N.(2009). Multi-Temporal Soil Erosion Risk Assess-ment in N . Chalkidiki Using a Modified Usle Ras-ter Model. EARSeL eProceedings, 8 (5), 40–52.

Gobin, A., Jones, R., Kirkby, M., Campling, P., Govers,G., Kosmas, C. & Gentile, A. R. (2004). Indicatorsfor pan-European assessment and monitoring of

soil erosion by water. Environ. Sci. Policy, 7 (1),25–38.

Government of Malta. (2002). Malta National Report,Submitted by the Government of Malta to theWorld Summit on Sustainable Development. Gov-ernment of Malta. Malta.

Grove, A. T. & Rackham, O. (2001). The Natureof Mediterranean Europe: An Ecological History.Yale: Yale University Press.

Iraldo, F., Meli, A., Sacco, A., DeBrincat, R., Tam-burini, F., Bertoneri, F. & Rocchi, F. (2013). De-velopment of environmental monitoring strategyand environment monitoring baseline surveys.Malta Environmental and Planning Agency. Flori-ana.

Le Bissonnais, Y., Montier, C., Jamagne, M., Daroussin,J. & King, D. (2002). Mapping erosion risk forcultivated soil in France. CATENA, 46 (2-3), 207–220.

Lee, C. R. & Skogerboe, J. G. (1985). Quantification oferosion control by vegetation on problem soils. InS. A. El-Swaify & W. C. Moldenhauer (Eds.), Soileros. conserv. (pp. 437–444). Ankeny: Soil Conser-vation Society of America.

Malısek, A. (1992). Optimum slope length in relation towater erosion (in Slovak). Ved. prace Vyskumnehoustavu podnej urodnosti, 17, 203–220.

MALta Soil Information System. (2003). MALSIS, ASoil Information System For The Maltese IslandsLIFE 00/TCY/MT/000036 Report, Malta. Min-istry of Agriculture and Fisheries.

Morgan, R. P. C. (1995). Soil erosion and Conservation.Edinburgh: Longman.

National Statistics Office. (2003). Environmental protec-tion expenditure by local councils. NSO.

Naveh, Z. (1991). A model of multiple-use managementstrategies of marginal and untellable Mediter-ranean upland ecosystems. In J. Cairns, G. P. Patil& W. E. Waters (Eds.), Environ. biomonitoring,assessment, predict. manag. - certain case stud.relat. quant. issues. (pp. 269–286). Maryland,USA: International Co-operative Pub. House.

Novotny, V. & Olem, H. (1994). Water Quality: Preven-tion, Identification, and Management of DiffusePollution. Hoboken, NJ: Wiley.

Oldeman, L. R., Hakkeling, R. T. A. & Sombroek, W. G.(1991). GLASOD World Map of the Status ofHuman-induced Soil Degradation (S. InternationalSoil Reference and Information Centre, UnitedNations Environmental Programme. Nairobi.

Pasak, V., Janecek, M. & Sabata, M. (1983). Ag-ricultural land protection against erosion (inCzech), Metodiky pro zavadenı vysledkuvyzkumu

10.7423/XJENZA.2015.1.06 www.xjenza.org

Sultana, D. (2015).Xjenza Online, 3:41–50. 50

do zemedelske praxe. Ustav VedeckotechnickychInf. pro Zemedelstvı, 11, 77.

Pinto Correia, T. (1993). Land abandonment: Changesin the land use patterns around the Mediterraneanbasin. CIHEAM, 1 (2), 97–112.

Price, S. & Nixon, L. (2005). Ancient Greek AgriculturalTerraces: Evidence from Texts and ArchaeologicalSurvey. Am. J. Archaeol. 109 (4), 665–694.

Renard, K. G., Foster, G. R., Weesies, G. A., Mc-Cool, D. K. & Yoder, D. C. (1997). Predicting soilerosion by water: a guide to conservation plan-ning with the Revised Universal Soil Loss Equa-tion (RUSLE). In Agric. handb. no. 703 (pp. 1–384). Agricultural Research Service: Washington,DC.

Renard, K. G. & Freimund, J. (1994). Using monthyprecipitation data to estimate R-factor in the re-vised USLE. J. Hydrol. 157, 287–306.

Renschler, C. S., Mannaerts, C. & Diekkruger, B.(1999). Evaluating spatial and temporal variabil-ity in soil erosion risk - Rainfall erosivity and soilloss ratios in Andalusia, Spain. Catena, 34 (3-4),209–225.

Tanti, C., Role, A., Borg, C. & Calleja, I. (2002). Pro-tection of Soil and Rural Landscapes in NorthwestMalta. In Unep/map (united nations environ. pro-gram. action plan). 2003. map camp proj. “malta”final integr. proj. doc. sel. themat. doc. Athens:MAP Technical Reports, Series No 138, Volume2.

Whitelaw, T. M. & French, C. A. I. (1999). Soil erosion,agricultural terracing and site formation processesat Markiani, Amorgos, Greece: The micromorpho-logical perspective. Geoarchaeology, 14 (2), 151–189.

Wischmeier, W. H. (1959). A rainfall erosivity index fora Universal Soil Loss Equation. Soil Sci. Soc. Am.J. 23 (3), 246–249.

Wischmeier, W. H. & Smith, D. D. (1958). Rainfall en-ergy and its relation to soil loss. Trans. Am. Geo-phys. Union, (39), 285–291.

Wischmeier, W. H. & Smith, D. D. (1978). Predict-ing rainfall erosion losses: a guide to conserva-tion planning. In U.s. dep. agric. handb. no. 537(pp. 1–69).

10.7423/XJENZA.2015.1.06 www.xjenza.org

Masini, J. and Francalanza, A. (2015).Xjenza Online, 3:51–55.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.07

Research Article

Typing Actors using Behavioural Types

J. Masini, A. FrancalanzaDepartment of Computer Science, University of Malta, Msida, Malta

Abstract. The actor model of computation assists and dis-ciplines the development of concurrent programs by for-cing the software engineer to reason about high-level con-currency abstractions. While this leads to a better handlingof concurrency-related issues, the model itself does not ex-clude erratic program behaviours. In this paper we considerthe actor model and investigate a type-based static analysisto identify actor systems which may behave erraticly dur-ing runtime. We consider the notion of behavioural typesand consider issues related to the nature of the actor modelincluding non-determinism, multi-party communication, dy-namic actor spawning, non-finite computation and a possiblychanging communication topology, which we contrast withexisting works.

1 IntroductionThe actor model (Hewitt, Bishop & Steiger, 1973) is be-coming increasingly prevalent in the development of highly-concurrent systems and constitutes the underlying modelof several mainstream technologies including programminglanguages such as Erlang (Armstrong, 2007; Cesarini &Thompson, 2009) and Scala (Odersky, Spoon & Venners,2011), and frameworks such as AKKA (“AKKA”, 2015) andCloud-Haskell (Haskell, 2015). In particular, this program-ming model aides and disciplines the development of concur-rent systems, facilitating the handling of concurrency-relatedprogramming issues i.e., race conditions, dead/livelocks andstarvation, whilst shielding from intricacies that can easilylead to errors (Haller & Sommers, 2012). Computation in theactor model is carried out by a number of single-threaded en-tities called actors executing concurrently within their ownlocal memory. The absence of shared memory forces act-ors to interact solely by means of asynchronous message-passing. In Figure 1 we show the basic structure of an actor,composed of three distinct elements i) a unique actor nameused (as an address) for communication; ii) an expressionwhich describes the actor’s behaviour; and iii) a mailbox

that stores received messages in order (of time of arrival).Whereas each actor may be sent messages from several act-ors at the same time (multi-participation), input may onlybe done from the actor’s own mailbox, unlike channel-basedmessage passing (Haller & Sommers, 2012).

i e q

Figure 1: An actor.

The actor also have the ability to dynamically change thetopology of their system by spawning additional actors dur-ing runtime. Communication topology may also be changedby the delegation of actor names in message parameters, al-lowing output to previously unknown actors. Actor compu-tation may also be non-terminating in order to implementfunctionality such as that of servers or web-services. Dueto the concurrent nature of the programming model, multi-participation and asynchrony, the messages received in anactor’s mailbox may be different for each execution, causingnon-deterministic actor behaviour. As a result, actor-modelinspired technologies such as Erlang and AKKA extend in-put assisted with a pattern-matching mechanism (Armstrong,2007; Cesarini & Thompson, 2009; Odersky et al., 2011),which allows messages to be retrieved from the mailbox in anorder other than the one received in. We emphasise that ourpresentation of the actor model is derived from a high leveldescription as described by Agha (1986), Clinger (1981) andinstantiations of this model within programming languagessuch as Erlang (Armstrong, 2007; Cesarini & Thompson,2009).

The rest of this paper highlights the actor model of compu-tation (Section 2) followed by an analysis of the issues (andrelated existing works) to adapt behaviour types for actors,(Section 3). We conclude by analysing related work in beha-vioural types that concern actors specifically and close withsome final remarks (Section 4).

Correspondence to: J. Masini (joseph.masini09@um.edu.mt)

c© 2015 Xjenza Online

Masini, J. and Francalanza, A. (2015).Xjenza Online, 3:51–55. 52

cl1

cl2

sr db〈req, cl1〉

〈req, cl2〉

〈val, cl1, sr〉

〈ack, true〉

〈rep, true〉

Figure 2: Banking system.

2 Actor Systems

Consider the actor system depicted in Figure 2 consisting oftwo clients, cl1 and cl2, which require credential validationfrom a server, sr, where the latter in turn requestions the as-sistance of a database service, db. Each actor is identifiedby its unique actor name, and output operations are represen-ted as directed lines with messages as labels (and we abstractaway from the actors’ expression and mailbox structures).

In Figure 2, we show the validation process for one cli-ent; the client sends a request to the server in the form of atuple tagged by the label req containing its own address forvalidation purposes, shown as 〈req, cl1〉 (for cl1and similarly〈req, cl2〉 for cl2). Labels are employed to tag distinct mes-sages for selective input from an actor’s mailbox (throughpattern-matching). Upon input, the server processes the cli-ent’s request by extracting the client’s address and createsa new message tagged with the label val containing the cli-ent’s address and its own (as a reply address), 〈val, cl1, sr〉.The server sends the latter message to the database serviceand awaits a reply from the server. After the database ser-vice inputs the server’s request, validation is carried on theclient’s name and the result is sent to the requester’s address(in this case the server) in the form of 〈ack, true〉 specifyingthe validation result. Once the server retrieves the message,it creates a new message, 〈rep, true〉 and sends it to the clientcurrently being handled. It is important to note that the serveris initially unaware of the client(s). Since it acts at the inter-mediary between it and the database service (which is awareof the client in order to verify its identity), when the serverinputs this message it is temporarily made aware of the clientby the address in its request in order to carry out the val-idation task with the database service and reply back to the

client with the result. Yet the system specified in Figure 2may run into problems; the server represents a servicing bot-tleneck since it serializes every request and temporarily haltsservicing other clients.

Consider an improved arrangement of our banking systemshown in Figure 3; instead of handling the task locally, theserver assigns the service request to a new actor which itspawns acting as a task handler, th. This allows the serverto remain responsive to other client requests by removing thecomputational load of redirecting messages back to the re-spective clients.

In this new arrangement we use the actor’s ability to spawnadditional actors during runtime and have the spawned taskhandle credential validation (of that particular client) withthe database handler and subsequently complete the servicerequest interaction with the client. This is achieved by thenewly spawned task handler delegating its address instead ofthe server’s in the request to the database service, 〈val, cl, th〉.It is important to note that the client is unaware that the ser-vice request was completed by the task handler instead of theserver.

As both banking examples show, the actor model discip-lines the implementation of a concurrent system; it forcesthe software engineer to avoid the mechanisms (and possiblepitfalls) of shared memory by abstracting reasoning on theprocessing components as distinct entities executing withintheir own environment. In spite of this, the actor model doesnot guarantee the absence of erratic behaviour during the ex-ecution of such a system. In particular, the aforementionedsystems are still susceptible to execution errors i.e., systemcrash. Consider a client which sends a boolean value insteadof its address for validation, as shown in Figure 4. When thedatabase service attempts validation of the client credentials

cl1

cl2

sr th db〈req, cl1〉

〈req, cl2〉

spawn

〈val, cl1, th〉

〈ack, true〉〈rep, true〉

Figure 3: Improved bank system.

10.7423/XJENZA.2015.1.07 www.xjenza.org

53 Masini, J. and Francalanza, A. (2015).Xjenza Online, 3:51–55.

clerr sr th db〈req, true〉 spawn 〈val, true, th〉

Figure 4: Erratic client 1.

(not shown in the figure), it will require an address but isprovided with a boolean value, thereby crashing and haltingthe database service.

In Figure 5 we show another type of erratic behaviourwhere the database service would receive a genuine request,but never reply back. In such cases, the said system shouldbe analysed against a form of liveness property to discardsystems which get stuck.

We would like to formalise a static type system (Pierce,2002; Cardelli, 2004) for the actor model, which would guar-antee the absence of a select set of errors in actor systems.However, the development of a type-based static analysis foran actor system is non-trivial due to a number of reasons. Inorder to verify each actor’s behaviour, we require to analysethe mailbox which essentially defines the actor’s behaviour(Agha, 1986). The messages and the order in which theyare received are the main variant on how an actor executesupon input e.g., in our banking example, the server’s nextoperation is dictated by the messages inputted from its mail-box, which in turn depends on the current state of its mail-box. However, since actor mailboxes may receive differenttypes of messages, traditional type systems for concurrencyare inapplicable, as these assume invariance with respect tothe values communicated (Marlow & Wadler, 1997).

3 Typing ActorsBehavioural types (compared to traditional types),allow the analysis of how computation occurs byproviding notions of causality and choice (“Beha-vioural Types for Reliable Large-Scale SoftwareSystems - The Foundations of Behavioural Types:http://www.operationalsemantics.net/behaviouralwiki”,n.d.). These have been extensively used to statisticallyanalyse a wide range of computing aspects, ranging fromensuring party compatibility in protocols (Mostrous &Vasconcelos, 2011) to correct method call ordering inobject-oriented systems (Gay, Vasconcelos, Ravara, Gesbert& Caldeira, 2010). Despite numerous existing works,proposed approaches are not directly applicable to analysethe actor systems. We aim to develop our own notationof behavioural types, whilst retaining the expressivity ofthe actor model. However, this is far from trivial as werequire to consider several issues including concurrencyand asynchrony, non-terminating behaviour, dynamic actor

spawning and name delegation.

Concurrency and Asynchrony Concurrency is one mainsource of complication; due to potential non-determinism,we are forced to consider all possible execution interleavingsamongst actors. Actor communication is also asynchron-ous, hence actors send messages irrespective of the executionstate of the target actor. Due to this combination, actors thatare sent messages by several other actors at one time (multi-participation) do not always guarantee the specific order ofmessages in the recipient’s mailbox. Recall the server in Fig-ure 3; client cl2 can send a message to the server while it ishandling a previous request from client cl1. At the same time,the server may receive a reply for the database service for therequest of cl1. Figure 6 shows three instances of this server’sstate described by its name, expression and mailbox contentsreceptively. Figure 6 (a) describes the server’s mailbox con-taining a request by the client, cl2, and an acknowledgementsent back by the database service for the current client re-quest being processed, cl1. The server’s mailbox may, fromthis state: i) increase in size with the reception of an addi-tional message, Figure 6 (b), say from another client, cl3; orii) decrease in size with the server inputting from its mail-box, Figure 6 (c). In the latter case, since the server requiresto process two different forms of messages (specifically cli-ent requests in the form of 〈req, client name〉 and databaseservice replies, 〈ack, boolean〉) the server employs pattern-matching to extract messages other than the order presentin the mailbox, which complicates our static analysis of themailbox for each actor.

(a)

(b)

(c)

sr

sr

sr

. . . 〈req, cl2〉, 〈ack, true, cl1〉

. . . 〈req, cl2〉, 〈ack, true, cl1〉, 〈req, cl3〉

. . . 〈req, cl2〉

Figure 6: Mailbox communication structure.

Existing works, such as the work done by Honda, Yoshidaand Carbone (2008), extend the notion of (binary) sessiontypes to a multi-party setting over a π-calculus with asyn-chronous communication semantics. In contrast, communic-ation is carried by means of channels which may be used

cl sr th db〈req, cl〉 spawn 〈val, cl, th〉

Figure 5: Erratic client 2.

10.7423/XJENZA.2015.1.07 www.xjenza.org

Masini, J. and Francalanza, A. (2015).Xjenza Online, 3:51–55. 54

for input and output by several entities but only up to two atone time, while in the actor model, each mailbox may onlybe inputted from by the mailbox owner and may be output-ted by several entities at any time. Furthermore, input froma channel is on a first-in-first-out basis, unlike the pattern-matching input in an actor setting which allows input in adifferent order than from the one received, thereby havinglooser tolerances. They employ a notion of types inspiredfrom the paradigm of global programming - interaction typesare specified by a global description (or global types) of theoverall communication behaviour between peers and are sub-sequently projected to extract the individual endpoint localsessions of each peer. However, this requires the knowledgeof the communicating entities a priori, whereas in the actormodel the topology may change during runtime by the dy-namic spawning of actors. This is further aggravated bythe possibility of delegation, as the communication topo-logy may dynamically change to include actors unknown atruntime.

Non-terminating Behaviour We have to consider the abil-ity of actor-inspired technologies to specify non-finite com-putation (Armstrong, 2007; Cesarini & Thompson, 2009;Odersky et al., 2011). Similar to the server and databaseservice in our banking scenarios, actors can carry out a spe-cific operational sequence for an unbounded number of timese.g., the database service’s operation sequence to receive arequest for validation, process the credentials and send backthe acknowledgement. We have to ensure safety for eachpossible operational sequence, and with the addition of non-determinism, we have to consider all possible interleavingsfor each operational sequence e.g., it might be that in one ex-ecution, the request from one cl1 is processed by the serverfirst (and hence serviced by the database service), whereasin another execution cl2 might be processed first. Severalworks have tackled the problem of recursive behaviour (orsome form of repetition or replication of behaviour) (Hondaet al., 2008; Caires & Vieira, 2010), however in our case wehave to consider non-termination as well.

Dynamic Actor Spawning The actors’ ability to spawnadditional actors during runtime (Agha, 1986) is another is-sue that complicates the development of our static analysis.Our type system must support actor systems that may dy-namically change the number of participants, similar to ourbank system in Figure 3, whilst ensuring that these actorsare still safe when interacting within the actor system. Thisis substantially more complex when it is coupled with pos-sibly non-terminating computation. It requires assure safetyin systems with (possibly) an unanticipated number of par-ticipants which may only be known at runtime e.g., in ourbanking system, since we may not know a priori the numberof clients that are going to interact with our system, we can-not know the number of actors spawned by the server. Anextension of the work done by Honda et al. (2008) handlesthe issues of dynamic participation by the parametrization of

sessions according to the number of participants (Denielou,Yoshida, Bejleri & Hu, 2012). However, the abstractions re-quired for the communication constructs are rather differentfrom the ones in the actor model, as input and external choiceare modelled separately and messages in channels retain thereceived ordering. Caires et al. present a novel notion ofbehavioural types referred to as conversation types (Caires& Vieira, 2010) to address multi-party interaction with dy-namic interaction. The latter are similar to those found inservice-oriented computing, which are conceptually analog-ous to the communication interactions in the actor model.The type system is based on an extension of the π-calculusthat addresses the issue of non-deterministic communicationby labelling output and input messages, similar to the com-munication primitives in our actor calculus. In contrast, theactor mailbox is more dynamic from the channels employedin the conversation calculus (Vieira, Caires & Seco, 2008),whereby our input operation allows (a pattern-matched) se-lection other than the first value found in the channel. Thetype system employed in (Caires & Vieira, 2010) uses a gen-eralisation of session types adapted to multi-party interac-tions. This is achieved through the merging of local andglobal types which are used to compositionally distributeparts of the protocol between a number of participants, someof which may be unknown at runtime.

3.1 Name DelegationThe ability to delegate a name by sending a copy of theactor’s address as a message parameter allows dynamic chan-ging of the communication topology. While beneficial andallows communication to previously unknown actors, it de-centralises control and introduces further process interleav-ings e.g., the server in Figure 2 is previously unaware ofthe clients, until they send a request specifying their ad-dress. This is additionally complicated with the possibilityof dynamic actor spawning, as we require to analyse pos-sible communication paths between actors which are onlyknown at runtime, e.g., the task handler is spawned duringruntime in Figure 3, and the database service is made awareof it from the task handler’s request. Combined with non-terminating behaviour, this increases the complexity of thepossible communication topology, as the number of parti-cipants and their interactions is unknown before runtime. Asproposed by Caires and Vieira (2010), conversation types al-low the type-checking of dynamic conversations, where aparticular slice of the conversation is delegated to an otherparticipant. Actors may also dynamically send addresses toenable interaction with possibly, previously unknown act-ors. Caires and Vieira (2010) enable dynamic participationby modelling multiparty conversation through name passing.This may be contrasted with the approach taken by Honda etal. (2008), where the authors distinguish between the passingof values and the passing of session to another entity at thelevel of the calculus. In our case, delegation is more subtle as

10.7423/XJENZA.2015.1.07 www.xjenza.org

55 Masini, J. and Francalanza, A. (2015).Xjenza Online, 3:51–55.

we communicate actor names as values, where some of theactor names may even be unknown before runtime.

4 ConclusionThere has been limited work in the area of behaviouraltypes involving the actor model specifically (Crafa, 2012;Mostrous & Vasconcelos, 2011). Mostrous and Vasconcelos(2011) propose type system for a featherweight Erlang calcu-lus, which lacks notions of internal choice and infinite com-putation. They analyse actor systems to eliminate possibleactor impersonation which may cause malicious behaviour.However this is not our current goal as it enforces a specificprogramming approach thereby limiting actor (language) ex-pressivity. Mostrous and Vasconcelos (2011) employs thenotion of session types to model the protocol of specific se-quences and forms of messages. This is used to ensure thatan actor handles all the messages in its mailbox, and receivesall expected messages. However, this goes against our notionof the actor model as i) actors are not hindered by extra mes-sages (virtually) inside their mailbox; and ii) the absence ofa message in the mailbox does not constitute ill-behaviour,especially when we are reasoning on non-finite computation.In fact, an actor that does not receive a particular messagesimply blocks waiting for the desired input, a mechanismintrinsic to the actor model. Crafa (2012) proposes a workcloser to what we study, specified over an alternate actor cal-culus inspired by Odersky et al. (2011). Their work lacksthe notion of internal choice and constructs to express someform of repetitive behaviour. Furthermore, we contrast heruse of input semantics with (Crafa, 2012), where the authorhandles communication non-determinism by specifying non-deterministic semantics for actor input, which does not re-flect the input specification of the actor model. Also, Crafa(2012) employs an approach inspired by conversation typeswhere the behavioural types define the protocols as a se-quence of messages, branching from external choice. In or-der for typechecking to occur, a path is marked on each pro-tocol describing each actor and the expected messages, andthe path markings are distributed amongst each actor com-positionally. This approach allows Crafa to handle dynamicactor spawning and ensure adherence of each actor’s protocolthroughout the entire actor system.

We give an in-depth study of the issues involved in the ad-aptation of behavioural types to model actor systems. Wealso contrasted them against the current literature for actors,however we found these to be rather restrictive. We con-cluded that current type systems are not flexible enough toallow analysis of the actor model and we consider it as apromising area for further research and study.

ReferencesAgha, G. (1986). An Overview of Actor Languages. In Proc.

1986 sigplan work. object-oriented program. (pp. 58–67). OOPWORK ’86. New York, NY, USA: ACM.

AKKA. (2015). Retrieved from http://www.akka.ioArmstrong, J. (2007). Programming Erlang - Software for a

Concurrent World. The Pragmatic Bookshelf.Behavioural Types for Reliable Large-Scale Software

Systems - The Foundations of Behavioural Types:http://www.operationalsemantics.net/behaviouralwiki.(nodate).

Caires, L. & Vieira, H. T. (2010). Conversation Types. Theor.Comput. Sci. 411(51-52), 4399–4440.

Cardelli, L. (2004). Type Systems. In Comput. sci. eng.handb. CRC Press.

Cesarini, F. & Thompson, S. (2009). ERLANG Programming(Media, Inc). O’Reilly.

Clinger, W. D. (1981). Foundations of Actor Semantics.Cambridge, MA, USA: Massachusetts Institute ofTechnology.

Crafa, S. (2012). Behavioural Types for Actor Systems.CoRR, abs/1206.1.

Denielou, P.-M., Yoshida, N., Bejleri, A. & Hu, R. (2012).Parameterised Multiparty Session Types. Log. Meth-ods Comput. Sci. 8(4).

Gay, S. J., Vasconcelos, V. T., Ravara, A., Gesbert, N. &Caldeira, A. Z. (2010). Modular Session Types for Dis-tributed Object-oriented Programming. SIGPLAN Not.45(1), 299–312.

Haller, P. & Sommers, F. (2012). Actors in Scala. USA: Ar-tima Incorporation.

Haskell, C. (2015). Cloud Haskell website: retrieved fromhttp : / / www . haskell . org / haskellwiki / Cloud %5C Haskell

Hewitt, C., Bishop, P. & Steiger, R. (1973). A universal mod-ular ACTOR formalism for artificial intelligence. In Ij-cai (pp. 235–245). Morgan Kaufmann.

Honda, K., Yoshida, N. & Carbone, M. (2008). MultipartyAsynchronous Session Types. SIGPLAN Not. 43(1),273–284.

Marlow, S. & Wadler, P. (1997). A Practical Subtyping Sys-tem For Erlang. In Proc. int. conf. funct. program. (icfp’97 (pp. 136–149). ACM Press.

Mostrous, D. & Vasconcelos, V. T. (2011). Session Typingfor a Featherweight Erlang. In Proc. 13th int. conf. co-ord. model. lang. (pp. 95–109). COORDINATION’11.Berlin, Heidelberg: Springer-Verlag.

Odersky, M., Spoon, L. & Venners, B. (2011). Programmingin Scala: A Comprehensive Step-by-Step Guide, 2NdEdition (2nd). USA: Artima Incorporation.

Pierce, B. C. (2002). Types and Programming Languages.Cambridge, MA, USA: MIT Press.

Vieira, H. T., Caires, L. & Seco, J. (2008). The Conversa-tion Calculus: A Model of Service-Oriented Compu-tation. In S. Drossopoulou (Ed.), Program. lang. syst.(Vol. 4960, pp. 269–283). Lecture Notes in ComputerScience. Springer Berlin Heidelberg.

10.7423/XJENZA.2015.1.07 www.xjenza.org

Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.08

Research Article

Assessing potential output growth of the Maltese economy using aproduction function approach

A. G. Grech, B. MicallefModelling and Research Office, Central Bank of Malta and London School of Economics and Political Science

Abstract. After outlining the various methods usedto estimate potential output, this article presents estim-ates for Malta, the smallest member of the euro area, de-rived from one of the most commonly used methods, theproduction function approach. Given the uncertaintysurrounding these kinds of estimates, these estimatesare compared with those made for Malta by other in-stitutions using different methods. Based on this ana-lysis and on a cross-country comparison with other euroarea economies, a number of observations are made thatwould enable potential growth to accelerate and resultin a faster economic convergence.

1 Introduction

Malta is the smallest member of the euro area, compris-ing less than 1 % of the monetary union’s economic size.It also has one of the lowest levels of GDP per capitaamong euro area members and is located in a region, theMediterranean, which has been characterised by polit-ical unrest and severe economic distress in recent years.To be able to surmount these difficulties and convergeto the higher relative economic standards of its fellowmonetary union members, Malta faces the challenge ofachieving relatively higher rates of growth in its poten-tial output. The latter is usually defined as the highestlevel of output achievable without generating inflation-ary pressures in factor markets.

Estimating potential output has always been a chal-lenge to policy makers, especially in economies undergo-ing structural changes (like Malta, which is moving froma manufacturing and tourism-based economy to a morediversified and higher value-added services based eco-nomy) or in the aftermath of wars, natural disasters orfinancial crises (Malta is surrounded in the north by eco-nomies embroiled in the sovereign debt crisis and in the

south by countries who have gone through the economicturmoil of the Arab Spring). Conceptually, potentialoutput is often defined as the sustainable productioncapabilities of an economy determined by the structureof production, the state of technology and available re-sources. A closely related concept is the output gap,defined as the difference between actual and potentialoutput. Measures of the output gap provide an indica-tion of the overall balance between demand and supplyconditions in the economy. This is generally considereduseful information by policy makers, for example to de-termine and predict price pressures, and to gauge pro-gress in fiscal consolidation.1

A number of studies have documented that financialand economic crises have a sizeable impact on the levelof potential output and that, following their occurrence,output does not revert to its previous growth trend butrather remains permanently below it.2 There are sev-eral factors that can affect the economy’s supply capa-city after a recession. On the production side, examplesinclude the scrapping of existing capital stock owing tobusiness failure, a slowdown in investment due to highuncertainty about future prospects and tight credit con-

1For a comprehensive review of the various policy applicationsof potential output measures, see, for instance, De Masi (1997)

2For instance, European Commission (2009) identifies threepossible scenarios in which the crisis could affect potential out-put. All three scenarios assume a short-run decline in the level ofpotential output but differ in terms of its long-run impact on thegrowth rate. In the first scenario, potential output growth accel-erates after the crisis, therefore allowing the economy to recoverall lost output and hence, the level of potential output returnsto its initial trajectory. The second scenario assumes that poten-tial growth rate returns to its pre-crisis rate in the long-run butthere is a permanent downward shift in the level of potential out-put. The third scenario assumes that the growth rate of potentialoutput will be permanently lower after a crisis, which implies anincreasing loss of potential output over time compared with thepre-crisis level.

Correspondence to: A. G. Grech (grechga@centralbankmalta.org)

c© 2015 Xjenza Online

57 Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62.

ditions to firms. This state of events can, in turn, de-press the growth rate of total factor productivity, es-pecially if it leads to lower spending on research anddevelopment. On the labour market front, in additionto the erosion of skills, some workers who lose their jobsmay become discouraged in finding alternative work andleave the labour force entirely.

The implications of the global recession that wastriggered by the financial crisis of 2009 for the growthrate of potential output and whether the world (and inturn, the Maltese) economy will settle on a lower growthpath are still open issues. Apart from the crisis, demo-graphic developments will adversely affect potential out-put growth in a number of countries in the coming yearsowing to the shrinkage of the workforce because of anageing population. The assessment of such effects on thegrowth path of medium-term potential output remainsa key issue for economic policy analysis.

After outlining the various methods used to estim-ate potential output, this article presents estimates forMalta derived from one of the most commonly usedmethods, i.e. the production function approach. Giventhe uncertainty surrounding these kinds of estimates,they are compared with those made for Malta by otherinstitutions using different methods. Based on thisanalysis and on a cross-country comparison, a numberof policy recommendations and final observations aremade.

2 Methods

2.1 Alternative methods to estimate po-tential output

There are various methods available in the literatureto estimate potential output.3 These can be groupedinto three broad categories: the production functionor growth-accounting exercises, statistical approachesand measures computed from dynamic stochastic gen-eral equilibrium (DSGE) models. The first approach at-tempts to create an explicit model of the supply side ofthe economy using economic theory. The second simplyattempts to break down real GDP directly into a trendand a cyclical component. The third approach, althoughfounded in economic theory like the first, is conceptuallydifferent. It recreates the level of output of the economythat would prevail in the absence of the structural rigid-ities that form part of the underlying model.

The production function framework is generally con-sidered a useful way to explain the key economic forcesunderlying developments in output and growth in the

3For a non-technical overview of these alternative methods andpolicy implications, see Mishkin (2007) or Cotis, Elmeskov andMourougane (2005).

medium term.4 This approach provides a comprehens-ive framework for estimating potential output, with aclear link between output and its long-term fundamentaldeterminants. Thus, it may be used to assess the impacton potential output of structural changes and individualpolicies. Nevertheless, this approach has certain disad-vantages. First, it raises important data problems. Inparticular measures of capital stock are often not avail-able and data on hours worked are not very reliable.Second, it requires measures of the trend componentsof the factor inputs. Different assumptions about thesetrend components can lead to different estimates of po-tential output.

Statistical methods of estimating potential output arebased on extracting the trend from the output seriesusing statistical techniques. These methods can be di-vided into two categories. Univariate methods extractthe trend from the information contained in the out-put series in isolation, without using the informationcontained in other variables. A widely used approachin such a univariate estimation of potential output isthe Hodrick-Prescott (HP) filter. This filter extracts atrend component by trying to balance a good fit to theactual series with a certain degree of smoothness in thetrend. While such filters are relatively easy to use, theyhave a number of drawbacks, the most important be-ing the poor reliability of the end-of-sample estimatesand the arbitrary choice of the smoothing parameter.In addition, these methods take no account of economictheory or of information involving other series, whichmay help to separate the trend from the cycle. Thefilter will also smooth structural breaks, even if thesetake the form of clear shifts in the level or the growthrate of the series and, therefore, it generates mislead-ing estimates of potential output around these periods.Moreover, this simple filter is ill adapted to handle thehigh degree of volatility manifested in the time series ofvery small open economies (Grech, 2013).

Multivariate filters attempt to extract the trend us-ing the information in the output series in conjunctionwith information contained in other variables, most not-ably inflation or the unemployment rate. These tech-niques typically attempt to take into account empir-ical relationships, such as the Phillips curve or Okun’sLaw.5 They too, however, suffer from drawbacks, suchas the sensitivity of these estimates to the modelling spe-

4The European Commission, for instance, adopts this methodas it considers it is the only one that can “underpin a comprehens-ive economic assessment of the outlook” and is the most “satis-factory instrument to frame economic policy discussions or explainpolicy decisions to the public”. See Economic Policy Committee(2001). A sectoral production function approach is also used bythe Congressional Budget Office (CBO) in the US, details of whichare available in (Congressional Budget Office, 2001).

5See Micallef (2014) for an application of a multivariate filterfor Malta to estimate potential output and NAIRU.

10.7423/XJENZA.2015.1.08 www.xjenza.org

Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62. 58

cification of these relationships, including the treatmentof expectations and estimates of equilibrium concepts,like the non-accelerating inflation rate of unemployment(NAIRU) (Carnot, Koen & Tissot, 2005).

Finally, measures of potential output can also be de-rived using DSGE models. The latter are micro-foundedmodels in which certain rigidities, such as wage and pricestickiness, are used to match developments observed inmacroeconomic data. Removing these rigidities offersa natural way to define a measure of potential outputin a model-consistent way. This measure of potentialoutput is thus defined as the output level that would berealised in equilibrium if prices and wages were perfectlyflexible. The drawback of this approach is that, withoutrigidities, output – identified as “potential output” inthis framework – adjusts more rapidly than observed inthe data. This gives rise to more volatile measures ofpotential output than usually obtained by other meth-ods.6 Furthermore, DSGE estimates of potential outputare considered to be more model-dependent than the al-ternatives.

Given the advantages and drawbacks of various ap-proaches used in the literature, analysts do not usuallydepend exclusively on a single estimate of potential out-put. The most common approach is to rely on a pro-duction function approach and then compute alternat-ive estimates, most likely from a statistical model, as across-check. Disparities across potential output estim-ates and comparison with those published by interna-tional institutions, like the International Monetary Fund(IMF) and the European Commission, are often used asan indication of the uncertainty surrounding these es-timates. When available, survey data on the degree ofcapacity utilisation can also be used either as a checkon output gap estimates or as complementary informa-tion to inform policy makers on the current state of theeconomy.

2.2 The production function approachapplied for the Maltese economy

In line with the approach taken by international insti-tutions, in this article the benchmark approach is takento be a production function meant to model the sup-ply side of the local economy. This production functionrelates output to the level of technology and factor in-puts, namely labour and capital, by means of a constant-returns-to-scale Cobb-Douglas specification, namely:

Yt = TFPt ×Kαt × L

(1−α)t (1)

where Yt denotes output at time t, Lt the labour in-put, Kt the capital stock and TFPt the total factor

6For a discussion of alternative notions of potential output inDSGE models used in ESCB central banks, see (Vetlov, Hledik,Jonsson, Kuscera & Pisani, 2010).

productivity. TFP is derived as the HP-filter of the So-low residual, using the standard smoothing parameter of1600 for quarterly data. The Solow residual is that partof economic growth that cannot be explained throughgrowth in the capital stock or in the labour supply andis a proxy for productivity gains.

In the Maltese case the share of labour income is cal-ibrated at 0.58, based on the share of compensation ofemployees in gross value added (GVA) between 1995and 2012, adjusted for the proportion of the labourforce that consists of self-employed. The stock of cap-ital is calculated using the perpetual inventory method.7

Housing investment is excluded from the measurementof the capital stock. The capital stock thus includesboth public investment and non-housing investment bythe private sector.

In more detail, labour input comprises several keyvariables of the labour market:

Lt = WAPt × PRt × (1 − UNRt) ×HRSt (2)

where WAPt denotes the working-age population,PRt the trend participation rate, UNRt the NAIRU andHRSt the trend hours worked. To help derive potentiallabour utilisation, the trend participation rate and hoursworked are obtained using the HP filter. NAIRU is com-puted from a multivariate filter, which is based on well-established relationships in economic theory, such as thePhillips Curve and Okun’s Law Benes et al. (2010).

3 Estimates of potential outputgrowth for Malta

Chart 1 plots potential GDP growth and its decomposi-tion between 1995 and 2014, using the production func-tion approach.8 According to this method, Malta’s po-tential output growth declined from above 4 % in thelate 1990s to a trough of around 2 % in 2003, primar-ily owing to both a downward trend in productivityand a slowdown in investment, with the latter leadingto a slower accumulation of capital.9 Potential outputgrowth recovered slightly during the cyclical upswingbetween 2004 and 2008, peaking at around 2.6 % in2006. This increase was driven by a rebound in invest-ment and a stronger contribution from labour, due toa combination of favourable demographics, an increasein the participation rate and a decline in NAIRU. On

7The perpetual inventory method is based on the followingformula: Kt = (1−δ)Kt−1+It, where Kt is the level of the capitalstock, It is real investment and δ represents the depreciation rate.

8Estimates in this paper are based on ESA2010 methodology.Minor differences from the results presented in Central Bank ofMalta (2013) are due to the use of ESA1995 statistics.

9Similar results are reported in Grech (2004).

10.7423/XJENZA.2015.1.08 www.xjenza.org

59 Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62.

the contrary, the contribution from TFP maintained itsdownward trend.

The recession of 2009 had an adverse effect on po-tential output growth, which declined to around 1.8 %per annum between 2009 and 2010. The slowdown wasmainly due to a sharp contraction in the contribution ofcapital, while, that of TFP remained at very low levelsby historical standards. The contribution of capital topotential output growth declined from an average of 1.2percentage points in the cyclical upswing before the re-cession to around 0.6 percentage point between 2009and 2014. This was due to a slowdown in investment,with the investment-to-GDP ratio declining from a his-torical average of 22 % before 2009 to 18 % afterwards.The decline in investment was broad-based, affected notonly by contraction in the construction sector in recentyears, but also in other industrial categories, such asmachinery and equipment.

After reaching a trough in 2009, with a contribution ofless than 0.2 percentage points, TFP growth has beengradually rising after the crisis. The contribution ofTFP to potential output growth, however, is still a frac-tion of that experienced in the late 1990s and early2000s. From a longer-term perspective, the slowdownin TFP growth started before the crisis and could bea reflection of structural changes in the Maltese eco-nomy, such as the reallocation of resources to sectors,for instance, from manufacturing to services, where pro-ductivity growth is slower (Dabla-Norris et al., 2015).More generally, a striking feature of the evolution ofTFP in Malta is the absence of procyclicality – pro-ductivity, whether measured in terms of labour or totalfactor productivity, generally tends to rise during boomsand falls during recessions – that is usually observed inother economies (Basu & Fernald, 2001).

The contribution of labour to potential output growthincreased significantly in recent years (see Chart 2). Inthe years before the recession, the main source was anincrease in the working-age population and, to a lesser

extent, the downward trend in NAIRU. On the contrary,both the trend participation rate and hours worked con-tributed negatively. The latter coincides with the in-crease in part-time employment, which lowers the hoursworked per person. Since 2009, the unfavourable ef-fects of an ageing population started to weigh in, witha gradual decline in the contribution of the working-age population. However, these effects have been out-weighed by the rising participation rate. This wasmainly driven by the rise in the female participationrate, which went up by 9.8 percentage points since 2008,by far the largest increase among EU countries, thoughit remains relatively low at 50.2 % in 2013. Moreover,the slight increase in NAIRU during the recession provedto be temporary, as the rise in the domestic unemploy-ment rate started to be reversed as early as 2010.

4 Comparison with estimates byother institutions

As already mentioned, estimates of potential outputare surrounded by a considerable degree of uncertainty.This stems from a number of factors, such as the un-observable nature of this variable and the sensitivity ofthe results to the chosen method. The uncertainty sur-rounding the estimates of potential output is especiallypronounced during a period of structural change in theeconomy or after a recession, when it is difficult to dis-entangle transitory effects from permanent ones.

One way of quantifying this uncertainty is by com-paring the results of the production function approachwith alternative estimates, computed for instance froma statistical model, or with those published by inter-national organisations. This section compares the es-timates made in the previous section of Maltese po-tential output and the output gap with four alternat-ive estimates. These include a standard HP filter andthe estimates published by the Ministry of Finance andthe European Commission. Estimates by the European

10.7423/XJENZA.2015.1.08 www.xjenza.org

Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62. 60

Commission, which are based on a production functionapproach similar to the one presented in the previoussection, refer to those published in the 2014 AutumnEconomic Forecasts.10 Estimates of potential outputfrom DSGE models are not available for Malta.

Chart 3 plots a range of estimates of potential out-put growth around the estimates derived using the pro-duction function approach as outlined above. Despitesome disparities across the various point estimates ofpotential output, the different models point to a sim-ilar story. Potential output slowed down significantlybetween the late 1990s and the early years of the lastdecade. Moreover, despite the moderate increase inthe supply capacity during the cyclical upswing beforethe 2009 recession, potential output growth had not re-covered to the rates prevailing in the late 1990s. Theweakness in economic activity following the 2009 reces-sion led to a slowdown in the growth rate of potentialGDP, which however started to recover in recent years,with some estimates even pointing to growth rates in2014 that exceed those registered in the pre-crisis period.

Building upon the various estimates for potential out-put, Chart 4 plots the range of estimates for the out-put gap. Allowing for some degree of uncertainty inpoint estimates, all the indicators are broadly consist-ent in the analysis of the business cycle fluctuations ofthe Maltese economy over the past 15 years. Between1996 and 2014, there were two clear periods when theeconomy was operating above potential. The first onewas in 2000, driven in part by a large (temporary) in-

10For details of the Commission’s production function approach,see (D’Auria et al., 2010) and (Havik et al., 2014). The Commis-sion’s approach is very similar to the one used in this paper withtwo main differences: (i) the labour share in the Commission’sproduction function is assumed to be the same for all EU coun-tries whereas in our case, it was calculated from the data i.e. theaverage share of labour in Gross Value Added adjusted for theshare of self-employed for the period 1995-2012 and (ii) differ-ences in extracting the structural components of some inputs inthe production function, such as TFP and NAWRU.

crease in activity in Malta’s semi-conductor industry inthe context of the international technology boom. Thesecond one occurred between 2007 and 2008, before theglobal recession.

This recession, which led to a fall in domestic outputin 2009, also pushed output below its potential and ledto a negative output gap in the range of −1 % to −2 %.The subsequent recovery in economic activity led to arelatively quick correction of the output gap, especiallywhen compared to other EU economies. The outputgap worsened slightly in 2012 owing to the slowdown inGDP growth due to the sovereign debt crisis in Europebut, following two years of above average GDP growth,the gap between actual and potential GDP was broadlyin balance by 2014.

5 Cross-country comparisonswithin the European Union

The impact of the great recession on Malta’s potentialoutput growth was more muted than in the rest of theEuropean Union (EU). Chart 5 plots selected potentialgrowth rates across the EU between 1996 and 2014. Po-land, which is shown at the top of the range, had themost consistently positive potential output path in theEU, with growth averaging 4.1 % per year, while Italyhad the lowest, at just 0.7 % on average. Note, however,that in some countries, potential output has exhibitedboom and bust dynamics, with relatively high growthrates before the crisis, as in Ireland and Latvia, butshrinking potential since 2008. Italy and Poland have,in fact, been selected on the basis of the consistency andstability of their growth path, in order to ascertain themain contributing factors behind success or failure.

The relatively strong growth of potential output inPoland appears to be driven by a consistent declinein structural unemployment, supported by a relativelystable, albeit moderate investment to GDP ratio. Thelong-term decline in Italian growth potential, from

10.7423/XJENZA.2015.1.08 www.xjenza.org

61 Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62.

around 2.3 % in 1980s to less than 1 % in the 2000s,is driven by the downward trend in total factor pro-ductivity. Recent studies have attributed this slowdownto a combination of deteriorating competitiveness andresource misallocation, the latter most likely a resultof out-dated management practices and limited penet-ration of information and communication technologies(Hassan & Ottaviano, 2012).

Malta has been closer to the top of the range, espe-cially in the late 1990s and in the last two years, andhas outperformed the average for the euro area, exceptduring the period of economic restructuring that pre-ceded EU accession in 2004. This bodes well for thepace of real convergence of the Maltese economy. Thegap between the two economies has widened in recentyears given the divergent path in economic growth ex-perienced after the crisis. In particular, growth in theeuro area seems to have stabilized at a lower level after2009, although this picture masks considerable hetero-geneity in the constituent countries. On the contrary,potential growth in Malta has already reached and, ac-cording to some estimates, even surpassed, the growthrate experienced in the cyclical upswing before the crisis.

Finally, as the sizeable changes in Malta’s potentialgrowth rates show, the business cycle in Malta is morevolatile than in the euro area, as would be expected fora very small open economy, but, with the exception ofthe last few years, it is quite synchronised with that inthe rest of the monetary union.

6 ConclusionAs in other euro area countries, in Malta the impact ofa slowdown in population growth has so far been offsetby rising participation rates. In contrast, as in othercountries with a better-than-average potential outputpath, Malta has been spared the large rise in the struc-tural unemployment rate that has affected many euroarea economies. However, the Maltese economy has alsowitnessed a significant decline in the investment ratio,

which is towards the bottom of the range among EUcountries. Moreover, TFP growth has been on a de-clining trend, especially when seen from a longer termperspective and, despite the pick-up in recent years, itscontribution to potential output growth is only a frac-tion of that observed in the late 1990s.

To facilitate the recovery in potential output, it is cru-cial to create a better business environment and gener-ate the conditions to sustain more start-ups, while at-tracting new businesses to Malta. The ageing Maltesepopulation limits the possible increases in participationrates, so that potential output growth is unlikely to besustained only through higher labour inputs. That said,policymakers need to continue to put in place the rightincentives for more people to join and remain longer inthe labour force, while pursuing structural reforms to re-duce unemployment. Given the increased share of verycompetitive high value-added service activities, it is alsoessential to have further investment in education to im-prove the quality of human capital and increase labourmarket flexibility. This must be supported by adequateinvestment in the supporting physical, communicationsand information technology infrastructure of the coun-try. Finally, policymakers need to continue to pursuestructural reforms that lead to an improvement in pro-ductivity.

References

Basu, S. & Fernald, J. (2001). Why is productivity pro-cyclical? Why do we care? In C. R. Hulten, E. R.Dean & M. Harper (Eds.), New dev. product. anal.(Chap. 7). Chicago, U.S.: University of ChicagoPress.

Benes, J., Clinton, K., Garcia-Saltos, R., Johnson, M.,Laxton, D., Manchev, P. & Matheson, T. (2010).Estimating potential output with a multivariatefilter. Int. Monet. Fund Work. Pap. WP/10/285.

Carnot, N., Koen, V. & Tissot, B. (2005). EconomicForecasting. London: Palgrave Macmillan.

Central Bank of Malta. (2013). Assessing the SupplySide of the Maltese Economcy using a Produc-tion Function Approach (tech. rep. No. 4). CentralBank of Malta. Valletta, Malta.

Congressional Budget Office. (2001). CBO’s method forestimating potential output: an update. The Con-gress of the United States, Congressional BudgetOffice. Washington D.C.

Cotis, J. P., Elmeskov, J. & Mourougane, A. (2005).Estimates of potential output: Benefits and pitfallsfrom a policy perspective. In L. Reichlin (Ed.),Euro area bus. cycle stylized facts meas. issues.London: Centre for Economic Policy Research.

Dabla-Norris, E., Guo, S., Haksar, V., Kim, M., Koch-har, K., Wiseman, K. & Zdzienicka, A. (2015).

10.7423/XJENZA.2015.1.08 www.xjenza.org

Grech, A. G. and Micallef, B. (2015).Xjenza Online, 3:56–62. 62

The new normal: a sector-level perspective onproductivity trends in advanced economies. Int.Monet. Fund Staff Discuss. Notes, SDN/15/03.

D’Auria, F., Denis, C., Havik, K., McMorrow, K.,Planas, C., Raciborski, R., . . . Rossi, A. (2010).The production function methodology for calcu-lating potential growth rates and output gaps.Eur. Econ. Econ. Pap. 420.

De Masi, P. R. (1997). IMF estimates of potential out-put: Theory and practice. Int. Monet. Fund Work.Pap. 97/177.

Economic Policy Committee. (2001). Report on potentialoutput and the output gap. European Commission.Brussels.

European Commission. (2009). Impact of the currenteconomic and financial crisis on potential output.Eur. Econ. Occassional Pap. 49.

Grech, A. G. (2004). Estimating the output gap for theMaltese economy. Univ. Libr. Munich MPRA Pap.no.33663.

Grech, A. G. (2013). Adapting the Hodrick-Prescott fil-ter for very small open economies. Int. J. Econ.Financ. 5 (8), 39–53.

Hassan, F. & Ottaviano, G. (2012). Productivity inItaly: the great unlearning. VoxEU.

Havik, K., McMorrow, K., Orlandi, F., Planas, C.,Raciborski, R., Roger, W., . . . Vandermeulen,V. (2014). The production function methodologyfor calculating potential growth rates and outputgaps. Eur. Econ. Econ. Pap. 535.

Micallef, B. (2014). A multivariate filter to es-timate potential output and NAIRU for theMaltese economy. Cent. Bank Malta Work. Pap.WP/05/2014.

Mishkin, F. (2007). Estimating potential output. FederalReserve Bank of Dallas.

Vetlov, I., Hledik, T., Jonsson, M., Kuscera, H. & Pisani,M. (2010). Potential output in DSGE models. Eur.Cent. Bank Work. Pap. 1351.

10.7423/XJENZA.2015.1.08 www.xjenza.org

Paola, F. D. (2015).Xjenza Online, 3:63–68.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.09

Research Report

Digital Technologies for virtual recomposition. The case study ofSerpotta stuccoes

F. Di Paola1,21IEMEST, Istituto Euro Mediterraneo di Scienza e Tecnologia, Department of Communication, Interactive Graphicsand Augmented Reality, 90139 Palermo, ITALY2Department of Architecture, (DARCH) - University of Palermo, 90128 Palermo, ITALY, 389.4324760

Abstract. The matter that lies beneath the smoothand shining surface of stuccoes of the Serpotta fam-ily, who used to work in Sicily from 1670 to 1730, hasbeen thoroughly studied in previous papers, disclosingthe deep, even if empirical, knowledge of materials sci-ence that guided the artists in creating their master-works. In this work the attention is focused on the solidperspective and on the scenographic sculpture by Giac-omo Serpotta, who is acknowledged as the leading ex-ponent of the School. The study deals with some par-ticular works of the artist, the so-called “teatrini” (ToyTheater), made by him for the San Lorenzo Oratory inPalermo. On the basis of archive documents and pre-vious analogical photogrammetric plotting, integratedwith digital solutions and methodologies of computer-based technologies, the study investigates and interpretsthe geometric-formal genesis of the examined works ofart, until the prototyping of the whole scenic apparatus.

1 Introduction

In the field of diagnostics, conservation, protection andvalorisation applied to Cultural Heritage, it is essential,in order to develop an effective scientific investigationprocess, to promote the formation of interdisciplinarygroups with different professional competences, to test,define and propose specific methodologies and operativeprotocols. In the specific field of architectural surveyand restoration, the evolution of techniques and the useof non-invasive equipment is increasing, and numerousscientific and technological progresses realised throughprojects financed by research authorities, universities,companies and firms specialised in the sector are ob-served.

Focusing on our experience, the acquisition, pro-cessing and understanding of heterogeneous data extra-polated from the application of interdisciplinary specificinstrumental methods determined the crucial point ofthe methodological approach aimed to propose an oper-ative protocol in the specific field of protection, conser-vation and valorisation of Cultural Heritage.

This work presents the first results of a wider study,which is included within a National Research PRIN2010-2011, whose title is “Architectural Perspective: di-gital preservation, content access and analytics”. Theunit in Palermo is testing some instrumental hard-ware/software methods (owned by the Department ofArchitecture, DARCH and the Department of Civil En-gineering, DICIV, of the University of Salerno), in or-der to develop investigation paths aimed at observing,building, elaborating, managing and visualising three-dimensional models of important architectonical ex-amples of solid perspective in Giacomo Serpotta’s sceno-graphic sculpture.

2 Stuccoes form: the case studyof the tile “St. Francis clothesa poor man”

Here, we report the in-depth analysis conducted on the“plastic little theatres” of the Oratory of St. Lawrence,which are emblematic evidence of Serpotta’s stuccoworkmanship, admirably combining geometry, art andarchitecture. The architectural-compositional complex-ity of the subject matter of the research and the inter-esting scenographic-spatial expedients, implemented tocreate the micro-architectures of these incredible trun-cated pyramidal “boxes”, constituted the foundations

Correspondence to: F Di Paola (francescodipaola@iemest.eu)

c© 2015 Xjenza Online

Paola, F. D. (2015).Xjenza Online, 3:63–68. 64

for this survey.For more than 70 years, the Serpotta family has rep-

resented the peak of stuccoes art in Sicily between 17th

and 18th centuries. Although the progenitor of the fam-ily was Gaspare Serpotta, his son Giacomo is recognizedas the most refined and famous artist. The activity ofthe school proceeded with Giacomo’s son, Procopio, andthen with the son of the latter, Giovanni, but Giacomo’sexpressive refinement would remain unequaled.

In 1600, in Palermo, religious orders and “Compan-ies” used to build little buildings adjoining the broth-erhood churches, destining them as places of worshipfor the performance of spiritual exercises, sermons andother religious activities. These small rooms, knownas “oratories”, were initially conceived as a very simpleornamental apparatus and composed of one single-aisle chamber illuminated by large rectangular windows,without chapels and with an arch between the chamberand the chancel, the latter being generally square andcovered by a small dome.

The baroque style and the competitiveness amongPalermo “Companies” in adorning their oratories con-tributed to the transformation and reintroduction ofthese places, through rich and luxurious decorationsmeeting the standards of that time.

The simple architectural frameworks of the pilasterstrips, the windows edges, the ledges and the bases wereadmirably enriched and livened up by complex decor-ative compositions constituting plasters representing al-legorical statues, little angels, garlands and festoons.

The figure of Giacomo Serpotta fitted this lively cul-tural mood well. The sculptor, with his brilliant flairand his abilities in the art of stucco, modelled and moul-ded unique sculptures.

Following the devastating and destructive events ofthe Second World War, many churches and oratoriesdecorated with Serpotta stuccoes were heavily damagedand, after restoration interventions, fell into decades ofoblivion. All this brought some decorative apparatus toa condition of deterioration about the point of no re-turn. Fortunately, at the end of 20th century, the entirecycle of stuccoes obtained the right attention, thanksfirst to the impassioned interest of D. Garstang, andfinally to the so called “Serpotta Project” initiated in1999 by the Soprintendenza BB CC (Monuments andFine Arts Office) of Palermo, which faced in its firststage the restoration of the decorative apparatus of tenmonumental buildings (Garstang, 1990). The “SerpottaProject” represented a unique opportunity to conduct asystematic sampling of stuccoes and to carry out a com-parative analysis of the results of physical chemical in-vestigation, pointing out new elements of an extraordin-ary artistic heritage that until that moment was studiedmostly in its historic and artistic aspects.

Unfortunately, the preservation interventions cannotcompletely restore the original reading of those partsof the decorative apparatus that were stolen. The at-tention of thieves was focused on those elements whichcould easily be sold on the antique black market.

Our interest focuses on the sculptor’s meticulous real-isation of the eight perspective scenes, made in littlemagical boxes which reveal his deep knowledge of thegeometrical laws governing vision and perspective per-ception.

These frames, placed at a height of almost 4.50 metresfrom the floor, contain some low relief hagiographicscenes of the two saints with figures in the round an-imating the plastic little theatre, appearing off the scalecompared with the big allegorical figures laid out on theedges (Palazzotto, 2004; Basile, 1981; Carandente, 1966;Argan, 1957).

As repeatedly pointed out by scholars, Serpotta suc-ceeds in moulding an innovative and spontaneous sceno-graphic space, relating natural and architectural ele-ments with human figures in the round, which do not al-ways meet perspective accuracy (undoubtedly, he differsfrom illustrious predecessors like Gagini and Brunelles-chi).

In this study we report the experience conducted onone of the eight Serpotta little theatres, particularly fo-cusing on the scene of “St. Francis clothes a poor man”(Montaperto, 2013).

The process of knowledge through the integration ofthe different survey methods has been performed on thesampled tile as an experimental test to structure an op-erative protocol to be extended to the study of all theoratory tiles.

3 The 3D Scanning Technique

In the last decade of the 20th century, together withthe development of the equipment (generally differenti-ated depending on the depth of the survey field, fromcentimetres to hundreds of metres, or from spatial resol-ution of the 3D model points generated with the meas-urement), the development of dedicated software hasoccurred. This has allowed computer acquisition andelaboration of three-dimensional models with mesh tri-angulate surfaces or NURBS (3D imaging) and the in-spection and analysis of the latter, aimed at the creationof digital archives, the reconstruction of complex scener-ies of virtual realities (especially in the field of archae-ology) and the realisation of copies through prototypingtechniques.

The use of these tools, employed by these three-dimensional metric measurement techniques, turned outto be suitable for the geometrical characteristics of thesubject matter of the research, boasting remarkable pro-ductivity and high precision standards when compared

10.7423/XJENZA.2015.1.09 www.xjenza.org

65 Paola, F. D. (2015).Xjenza Online, 3:63–68.

to the equipment of the past. With minimal non-invasive interventions, the survey maintained the en-tirety and the authenticity of the historical-artistic her-itage, highlighting the undeniable benefits of the digitalprocess in terms of reliability and management, determ-ining a data bank which is consultable and implement-able in relation to conservation and future transmission(Arbace et al., 2013).

Among the new technologies currently proposed forthe application to Cultural Heritage, the use of the 3Dscanner represents a significant example of how origin-ally far apart fields, such as the one of conservation, thatof research and that of advanced technologies industry,can find a common interest ground for the non-invasiveexperimental use of methodologies and innovative toolsfor analysis procedures of geometric-dimensional data,restoration and structural monitoring (Santagati, Inzer-illo & Di Paola, 2013; Di Paola, 2007).

The 3D scanning technique, as known, is based onthe distance measuring method through electromagneticwaves, also known as LIDAR (light detection and ran-ging). These specific tools generally use laser diasti-meters which measure the signal flight time, that is im-pulses diastimeters.

The combination of a diastimeter with these charac-teristics, with a set of high-precision mechanical appar-atuses, made the realisation of scanner laser sensors pos-sible. The tool mechanics allows the materialisation ofan acquisition direction, while the laser diastimeter ac-quires a distance along the direction itself. The result ofthe acquisition is a set of three-dimensional coordinatesin a reference system associated with the tool and re-ferred to a high number of points, which are hit by thelaser beam and belonging to a physical surface of the ob-ject of the survey. The set of these points is commonlycalled a “points cloud”.

As extensively described in literature (Serna et al.,2011; Crosilla & Dequal, 2006; Gruen, Remondino &Zhang, 2003), one of the logistical innovations of thesurvey process using this technology is in the acquisi-tion phase, which is defined by high informative dens-ity, due to the enormous amount of information auto-matically stored by the tool and the potentialities ofvisual and real time restitution of the object geometricform, even if in a dense set of points. The instrumentalmethod, in the first acquisition phase, limits the oper-ator to interact with the camera by defining only someparameters which influence the scan density and, on theother hand, makes the data processing much heavier inthe next post-processing phase. Postponing an in depth-analysis of the technique to other moments, we highlightthat there are different 3D scanners on the market, eachone with different characteristics in the acquisition prin-ciple, obtainable precision, acquisition range and speed.

For the drafting of the survey project conducted inthe oratory, we chose to use the 3D portable scanningsystem with structured light flash bulb Artec MH (in-strument of the Models Laboratory of the Department ofCivil Engineering DICIV of the University of Salerno).

Artec MH is a 3D scanner which utilises a rathersimple scanning procedure: it is sufficient to move unin-terruptedly around the object and film it from variousangles. Although the technical characteristics tell aboutan alleged irrelevance of the camera angle, it is easyto observe how rays, which are perpendicularly incid-ent and/or not tangent, assure a greater final accuracy(however working on surfaces that do not really reflect).The extremely versatile system (it does not need anymarker), which is functional, rapid and capable of ac-quiring almost 500.000 points per second, turned out tobe particularly suitable for the geometric-dimensionalcharacteristic of the object (indeed the acronym “MH”indicates a “Medium” application field, perfectly con-sistent with the dimensions of the plastic little theatres)and the goals to reach in terms of metric precision. Therelated proprietary software automatically joins, geore-ferencing them, all the acquired frames in a single mesh.The algorithm, in fact, recognises the geometry of theobject and, if the shape is “plastic” enough, as in thecase in question, it allows the correct alignment of thevarious captured 3D frames to visualise them in a singlemodel (therefore conserving the reference system), elim-inating the presence of holes and shadows due to backdrafts as much as is possible.

In particular, due to the specific shape of the ob-ject, the global alignment with a single shot strip-runwould not have been of immediate determination. Thatis the reason why a little-acquisitions project has beenimplemented, on the order of a few tens of seconds, asshown in the figure (repeatedly changing the originalshooting stations in order to capture data relating tothe parts which were initially hidden or in the shade).The intrinsic measuring speed of this equipment alloweda rapid acquisition, assuring a low relative error (about0.2 mm). The used tool, unlike the subsequent MHTmodel, does not permit the capturing of images and,therefore, in this case, the texturisation has been en-tirely performed during the post-processing.

The procedural phases of alignment, setting and tex-turisation, have been performed through the softwareGeomagic Studio. This is a reverse engineering softwareallowing the management of scanning or 3D photogram-metric shots data, points clouds processing, mesh mod-els generation, CAD surfaces design and exportationand the automatic creation of sections and/or NURBSsurfaces for modelling.

Without the automatic identification of the targets,the scanning alignment operations are managed through

10.7423/XJENZA.2015.1.09 www.xjenza.org

Paola, F. D. (2015).Xjenza Online, 3:63–68. 66

two functions: “manual setting” and “global setting”.The former makes a series of adjustments to the relat-ive spatial positions on the basis of couples of homo-logous points defined by the user (by means of surfacematching algorithms, particularly the ICP algorithm -Iterative Closest Point - which aligns the common partsof adjacent scans through the minimisation of the dis-tance between them). The latter automatically alignstwo or more scans according to the points arrangement.

The acquired data are then optimised through con-trols for the elimination of the marginal points andthe reduction of ground noise. In general, before thesubsequent editing operations, it is appropriate to im-plement an algorithm for noise reduction in order tobalance the systematic errors of the scanning phase(defined, indeed, as noise), thus reducing the points dis-persion for a more uniform distribution of the pointscloud. We speak about “local noise” when referring tothe scans before the alignment, or about “global noise”when associated with the final recorded model.

In the editing phase, it is possible to meld more pointsclouds in a single polygon model through merging opera-tions. Moreover, this tool allows the automatic uniformsampling of points and the final generation of surfaces(wrap). In order to improve the mesh quality differentapplications can be implemented, together with someoperations typical for points models, such as Sandpaperto remove local flaws, Fill Holes, which fills the datagaps, and so on. This last function highlighted the pres-ence of “gaps” and allowed the generation of patches,thus reconstructing the polygon mesh, according to dif-ferent methods relating to the surrounding conditions(Fill, Fill Partial, Create Bridges, Clean Up).

In the tile model, exported in .ply format within theCAD modelling software NURBS Rhinoceros, the di-gital statue of St. Francis has been placed in the originalposition (the modelling process of the statue is describedlater in section 4). The high dense mesh (more than7.7 million polygons) allows an integral reading of theperspective scene and an analysis of the scenographicsculptural composition as Serpotta conceived it.

4 The process of organic model-lation of the statue of St. Fran-cis

One of the predetermined goals is to fully reconfigure theperspective perception of the little theatres, reprodu-cing the missing figure sculptures virtually and in print,through the production of prototypes.

Here we present some first results of the virtual restor-ation performed on a sampled tile, “St. Francis clothesa poor man”, in which the statue of the saint is currentlynot present. As already underlined in the previous sec-

tions, the scientific methodological process of reproduc-tion of the sculpture has been made possible thanks tothe findings of graphic-numerical restitutions processedduring a previous stereo-photogrammetric survey con-ducted before the purloining of the statue. The 1:5 scaletables produced in the 80s and represented through con-tour lines, and the historical photographs found by oneof the few photographers allowed to enter the oratory,show the complexity of the organic form of the statue(D’Alessandro & Pizzurro, 1989).

Unfortunately, the indirect sources and the scaleddrawings report the volumetric-spatial information ofthe in the round statue, but are not exhaustive, as theyonly show its part in sight and the apparent contour.However, they are the only reliable scientific evidence ofthe complete scenographic apparatus.

In the field of Computer Graphics, the interactive di-gital sculpture techniques of numerical models representthe current status of the software evolution art in thefield of the organic modelling of free form complex sur-faces.

The tools implementation allows the user to interact-ively carve or paint 3D models enriching their geometricor superficial detail, showing potentialities measured interms of quality, precision and speed, in the drafting ofthe graphic compositions and the versatility of manage-ment and space control.

There is no doubt about the opportunity of explor-ation, contaminations, relations, measurements and in-formation, supplied by the continuous evolution of ex-peditive and automatic procedures in the use of variousproducts of the informatics era. In this lively experi-mentation, the progress of the culture of representationtools leads the experts to a specialised level of drawingtechniques knowledge, revealing increasingly stimulat-ing multidisciplinary application fields, especially in thefield of cultural heritage, valorisation and protection.

The panorama is rich and varied, so as to often gen-erate difficulties in choosing the more specific softwareaccording to personal needs and to require a high levelof applied knowledge (among the most popular in thefield of digital 3D sculpture there are the open sourcesoftware Sculptris and the paid software ZBrush andMudBox). This panorama offers different solutions ofCAD platforms with interfaces which maximise effect-iveness and flexibility of the work stream.

Dialog and library access boxes, navigation palettesrich in icons, and tree-structured drop-down menus al-low easy and more rapid web surfing and fast objectmanipulation through the addition, removal or manip-ulation of virtual clay layers, fostering the interactionamong different format files and an extraordinary con-trol of the detail figurative level.

With reference to the latter, which is linked to the

10.7423/XJENZA.2015.1.09 www.xjenza.org

67 Paola, F. D. (2015).Xjenza Online, 3:63–68.

variable control of the surface detail level, one of themost significant innovations in the field of contempor-ary computer graphics is the introduction of subdivi-sion surfaces (sub-d). These are very important, par-ticularly for the organic modelling of morphologicallycomplex models, because they allow the subdivision ofpolygons and edges of any mesh surface through spe-cific approximation algorithms (among the most fam-ous: Catmull-Clark, Loop, Butterfly, Kobbelt), turningit into another surface entirely made of quadrangularor triangular polygons. These polygons, taken together,form the control cage of the resulting polyhedral surface,which can easily be locally modified. This representa-tion technique had great applicability in the field of en-tertainment software design, but also in the architectureand industrial design fields.

We describe the manual modelling process which ledto the realisation of the prototype of St. Francis sculp-ture.

5 Conclusion

The fact-finding survey, divided into different in-depthanalysis phases, aims to test an operative protocol whichcould allow the interpretation and validation of someimportant aspects of the admirable analysed examples.The study examines the state of conservation of the ma-terial, the geometric-spatial shape and the constructivegenesis of the scenographic structure of the tiles, high-lighting the particular way the sculptor constructed thespace creating perceptual illusion effects.

This paper explains the procedural scientific path ap-plied to a sampled tile, but the in progress researchimplies that the methodology should be applied tothe whole sculptural heritage of the Oratory of St.Lawrence. The restoration fact-finding surveys and thedigital survey instrumental technologies formed the con-tent of an archive database rich in info-graphic inform-ation useful for future focused in-depth analyses. Thecreated three-dimensional models, represented in virtualreality through VRML language (Virtual Reality Mod-eling Language), are interactively and immersively ex-plorable.

Concerning the sculptures of the “plastic littletheatres”, eleven statues no longer exist (since the 90s),making the reading and the interpretation of the sculp-tural scenes depicting the lives of St. Francis and St.Lawrence difficult, and sometimes illegible. In this re-gard, we underline that, without the finding and theconsultation of the graphic composition of the previoussurvey campaign made by the team of the former De-partment of Representation of the Drawing Institute ofPalermo, it would not have been possible to assure ascientific approach in the virtual reconstruction of themodel of the St. Francis stolen statue.

As explained in the previous paragraphs, the three-dimensional printing procedure has already been star-ted, which will allow us to show to the users, even if notexpert, scaled models of the tile and the missing statue,reproducing an interpretative hypothesis of the overallvolume.

Among the future research goals, we plan to integ-rate the collected digital information with AugmentedReality (AR) technological systems. AR is a set of tech-nologies that allow us to “augment” a real scene. It isa new IT discipline, still in experimental stages, whichbelongs to “computer graphics” and involves the super-imposition of digital content on the observed real world.It combines new ICT and new forms of communication,showing an incremented reality representation in whichartificial/virtual sensorial information is overlaid to thenormal visualisation perceived through our senses.

The research team has already experienced the ex-traordinary potential of this new IT discipline in otherfields, recognising its special innovative educational-scientific communication, instruction and entertainingvalues offered by the used state-of-the-art technologies,which is particularly interesting in the field of CulturalHeritage (Di Paola & Inzerillo, 2011).

The development of an application dedicated to Ser-potta’s stuccoes, usable through the latest generationmobile devices (smartphones, android devices, iPads,iPods) will allow the visitor to access additional con-tent in the form of videos or 3D models, and to visual-ise the models of the virtual statues introduced in thereal perspective scenes. The “augmented” vision willenrich the guided visit with undeniable cultural bene-fits, completing the works of arts perception with enter-tainment multimedia content, pictures and historical in-formation, related itineraries, paths to be organised andeducational information about the; execution processes,stucco preparation and composition, the restoration andconsolidation works, the diagnostic investigations of thedecay and the state of conservation of the surfaces. Allthis is achieved in a non-invasive way, helping to pre-serve the building state of conservation for the futuregenerations.

Acknowledgments

We thank the Cultural Heritage Office of Archbishop’sSee of Palermo for having permitted us access to therooms of the Oratory of St. Lawrence. We thank Dr.Maria Luisa Montaperto of the Association “Amici deiMusei Siciliani” for her regular presence in the surveydays. A particular thanks to Dr. Angelo Mulone, Dir-ector of the qualified Laboratory for applied researchGeoLab of Carini (Palermo) for having lent useful equip-ment for the execution of the measuring operations.

We thank the digital sculptor Davide Di Lorenzo

10.7423/XJENZA.2015.1.09 www.xjenza.org

Paola, F. D. (2015).Xjenza Online, 3:63–68. 68

(http://davidedilorenzo7.wix.com/davidedilorenzo) forhaving actively collaborated to the organic modellingprocess of the St. Francis statue.

We thank also Arch. Matteo Fabbri of TryeCO 2.0(Ferrara company specialised in 3D printing service withany prototyping system) for having verified the model ofthe statue and prepared it for the 3D printing process.

ReferencesArbace, L., Sonnino, E., Callieri, M., Dellepiane, M.,

Fabbri, M., Iaccarino Idelson, A. & Scopigno, R.(2013). Innovative uses of 3D digital technologiesto assist the restoration of a fragmented terracottastatue. J. Cult. Herit. 14 (4), 332–345.

Argan, G. C. (1957). Il teatro plastico di Giacomo Ser-potta. Veltro, 29–33.

Basile, E. (1981). Le sculture e gli stucchi di GiacomoSerpotta (1st ed.). Palermo: Edizioni Librarie Si-ciliane.

Carandente, G. (1966). Giacomo Serpotta (1st ed.).Torino: Eri.

Crosilla, F. & Dequal, S. (2006). Laser scanning ter-restre (1st ed.). Udine: CISM.

D’Alessandro, M. & Pizzurro, P. (1989). La Prospet-tiva solida nella scultura scenografica del Serpotta.Rilievi fotogrammetrici con restituzione grafico-numerica. (1st ed.). Palermo: Arti grafiche Giord-ano.

Di Paola, F. (2007). Il Complesso monumentale delloSpasimo. Lettura e analisi attraverso il rilievo.(Doctoral dissertation, University of Palermo).

Di Paola, F. & Inzerillo, L. (2011). Augmented real-ity. The case of Salinas Museum of Palermo. InIx forum internazionale di stud. le vie dei merc.s.a.v.e. heritage. safeguard archit. vis. environ.herit. Naples.

Garstang, D. (1990). Giacomo Serpotta e gli stuccatoridi Palermo. Palermo: Sellerio Editore.

Gruen, A., Remondino, F. & Zhang, L. (2003). Image-based automated reconstruction of the greatbuddha of Bamiyan, Afghanistan. In Comput. vis.pattern recognit. work. Madison, Wisconsin, USA.

Montaperto, M. L. (2013). Oratorio di San Lorenzo(1st ed.). Palermo: Amici dei Musei Siciliani.

Palazzotto, P. (2004). Guida agli oratori. Confraternite,compagnie e congregazioni dal XVI al XIX secolo.Palermo: Kalos.

Santagati, C., Inzerillo, L. & Di Paola, F. (2013). Image-based modeling techniques for architectural herit-age 3d digitalization: limits and potentialities. Int.Arch. Photogramm. Remote Sens. Spat. Inf. Sci.555–560.

Serna, S. P., Scopigno, R., Doerr, M., Theodoridou, M.,Georgis, C., Ponchio, F. & Stork, A. (2011). 3D-centered Media Linking and Semantic Enrichmentthrough Integrated Searching, Browsing, Viewingand Annotating. In F. Niccolucci, M. Dellepiane,S. P. Serna, H. Rushmeier & L. V. Gool (Eds.),Vast int. symp. virtual reality, archaeol. intell.cult. herit. The Eurographics Association.

10.7423/XJENZA.2015.1.09 www.xjenza.org

Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: 10.7423/XJENZA.2015.1.10

Research Article

Architectural Perspectives in the Cathedral of Palermo: Image-BasedModeling for Cultural Heritage Understanding and Enhancement

L. Inzerillo1,3, C. Santagati2,31University of Palermo, Department of Architecture, viale delle Scienze, Edificio 8, 90128, Palermo2University of Catania, Department of Civil Engineering and Architecture, viale Andrea Doria n. 6, 95125, Catania3Euro Mediterranean Institute of Science and Technology (IEMEST), Via Emerico Amari, 123 - 90139 Palermo

Abstract. Palermo offers a repertoire of both artisticand architectural solid perspective of great beauty andin large quantity. This paper addresses the problem ofthe 3D survey of these works and their related studythrough the use of image-based modelling (IBM) tech-niques. We propose, as case studies, the use of IBMtechniques inside the Cathedral of Palermo. Indeed, thechurch houses a huge and rich sculptural repertoire, dat-ing back to 16th century, which constitutes a valid fieldof IBM techniques application.

The aim of this study is to demonstrate the effective-ness and potentiality of these techniques for geometricanalysis of sculptured works. Indeed, usually the surveyof these artworks is very difficult due the geometric com-plexity, typical of sculptured elements. In this study,we analysed cylindrical and planar geometries as well ascarrying out an application of perspective return.

1 Introduction

The architectural and artistic heritage concerning thesolid perspectives in Sicily is of inestimable value both inquantitative/qualitative typologies: antependiums, fres-coes, sculptures, tiles, theater, false vaults, and more.Already by the end of the 1400’s, architects and artistswere involved in the national problem of the embel-lishments of the interior spaces of both places of wor-ship, such as churches and oratories, and civic buildings.The greater mastery of the techniques of perspectivelaunched a new style in which geometry, art, dexterity,and architecture blended into a single amalgam of greatvisual and emotional effect; such as magic boxes thatreveal the mastery of the use of geometry in the ser-vice of perception perspective (Figure 1). Thus, a pris-

matic space is transformed into an aberrated pyramidalprism where narratives are represented, subjects charac-terizing the context (D’Alessandro, Inzerillo & Pizzurro,1983; D’Alessandro & Pizzurro, 1989; Inzerillo, 2004,2012).

Figure 1: The prismatic space in solid perspective

In Palermo, the applications of solid perspective rangefrom the majestic illusions of spherical domes to thepanels depicting the lives of saints. In this study, weanalyze the exemplars inside the Cathedral of Palermoin particular. This research aims to develop new aven-ues of investigation designed to detect, build, develop,manage and visualize three-dimensional models of im-portant architectural examples of the solid perspectivein spectacular sculpture. The goal is to study, developand transmit to future generations, through innovative,non-invasive and low cost methods, this cultural herit-age in a unique way that combines geometry, art andarchitecture.

In this research, the choice of the most significantcases was influenced by the inherent beauty and sophist-

Correspondence to: C. Santagati (cettina.santagati@dau.unict.it)

c© 2015 Xjenza Online

Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75. 70

ication of the work, its location within a production rain-fall in the area (authors who have created many works,which have been much investigated by historians, etc.),and the availability of the same work (possibility of ac-cess, permissions granted, heights compatible with theinstrumentation, etc.).

The quality and variety of works in solid perspectivein the Cathedral of Palermo provide a case study thatcan identify effective methods of investigation for ac-quisition through 3D image-based modeling techniquesthat can then be applied and verified in further studies.

The Cathedral of Palermo, and its art works, havealways been the subject of studies and insights for his-torians, archaeologists, art historians, lovers of sacredplaces, etc., in terms of artistic and sculptural aspects.Nevertheless, in spite of previous historical interpreta-tion conducted, the different art works – also those insolid perspective - have not been investigated with theaim of modeling and perspective rendering.

2 Repertoire of Solid Per-spectives in the Cathedralof Palermo

The Cathedral of Palermo is the result of historicalstratifications that have made this monument an ex-traordinary palimpsest of exceptional scientific value(Amato, 1728; Basile, 1938; Bellafiore, 1999; Zanca,1989). The building dates back to the Norman age, fromArchbishop Walter of the Mill in 1184. The first changesthat incisively modified the exterior, according to thetaste of the Aragonese-Catalan, took place between thefourteenth and the first half of the fifteenth century. Thefourteenth century additions concerned the great west-ern tower, the four corner towers, the main facade andmain entryway, the portal and the south of the sacristy,the windows of the aisles, the chapels and the sacristy.From the second half of the fifteenth century to the endof the sixteenth, the interior of the cathedral was re-modeled with fine moldings, altars, portals and severalsculptures by the Gaginiana school (Domenico, Anton-ello and his sons), Laurana and Spadafora.

In particular, the work that most changed the internalspace of the grandstand marble monument was built byAntonello Gagini and his sons between 1507 and 1574,which sheathed the apse and gave birth to a gradual“whitening” of the church interior 2.

The podium, with two orders and vertically articu-lated by pilasters, was composed of 75 sculptures, in-cluding sculpture in the round, and 14 episodes of thelives of the saints in high relief made of solid perspect-ive and placed at head height so that they were vis-ible to the visitors. This extraordinary podium was re-moved at the end of the eighteenth century as a result

Figure 2: Main nave of the Cathedral of Palermo

of the renovation of the Cathedral according to the pro-ject commissioned by Ferdinando Fuga (1761) and laterfinalized between 1781 and 1801 by Giuseppe Venan-zio Marvuglia, assisted by Salvatore Licata Attinelli andFrate Felice La Licata (Basile, 1926; Giuffre & Urbani,1993).

The surviving pieces of the podium were relocatedinside and outside the cathedral. Among these were14 panels relating to episodes of the lives of the saintsfound in the apse and the transept (Kruft, 1980; Rizzuti,2007; Vesco, 2011). They consist mostly of perspectivescenes built according to Brunelleschi style: a centralperspective with a stage set that through coffered ceil-ings, sidewalls and succession of vaulted arcades ampli-fies the depth of the environments Figures 3 u 4. Ori-ginally, the scenic locations of the panels were colored(blue and gold). Of some tiles (St. Philip, St. Matthewand St. Paul) we have the completion date (1527) (DiMarzo, 1880, 1884; Mancino, 2007).

Further surviving elements of the Gaginian podiumare located in the chapel of Mary of the Angels along thenorthern nave, flanked to that of the baptistery, whichhouses the statue of the Assumption of Mary and thebier of Mary. Below this, there are three solid perspect-ive scenes made in 1535 and of uncertain attribution,some indicate Antonello while others Fazio Gagini (Fig-ure 6).

In addition to the work of Antonello Gagini, inside thecathedral there are more panels in solid perspective suchas those made by his sons, Fazio and Vincent. Theseinclude such works as the marble arch of the chapel ofthe Crucifix between 1557 and 1565, although this workwas dismantled during the works which destroyed thepodium. Only a few panels with episodes of the Passionof Christ were reused and reassembled to form the altarof the Crucifix as it appears today (Figure 5).

The baptismal font (Figure 7(a)), made by Filippoand Gaetano Pennino in 1797, is next to the chapel ofMary of the Angels. The octagonal basin is supportedby a group of sculptures depicting the tree of knowledgeand Adam and Eve. Scenes of baptism according to the

10.7423/XJENZA.2015.1.10 www.xjenza.org

71 Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75.

Figure 3: Episodes of the lives of saints: “St. Philipand the Dragon”

Figure 4: “Jude Thaddeus of Edessa, who convertsking Abgar”

Figure 5: Altar of the crucifixion Figure 6: altar of Mary of the Angels

descriptions of the Acts of the Apostles are carved inbas-relief on the eight faces of the basin.

Other precious masterpieces include the two fonts setagainst the two pillars placed near the north and southentrances. The attribution of holy water placed nearthe southern entrance is uncertain (Figure 7(b)). DiMarzo (1880) attributes the work to Gagini (the fatherof Antonello), the Accascina (1959) to the choral workby Domenico Gagini, Pietro da Bonate and Gabrieledi Battista, and dates it to 1475-1480. Recent studiesattribute the font to Pietro da Bonate and date it to1464-1469 (Kruft, 1972).

Above the tank there are two lunettes, the lowerone showing the scene of Jesus’ baptism, and the up-per showing the blessing of a baptismal font. The fontwas completed by an octagonal canopy surmounted bya statue.

Some scholars date the basin on the opposite side (en-trance from the north portico) to 1553 and assume thatit was commissioned to the sculptors Giuseppe Spada-fora and Antonio Ferraro, also known as “Imbarracoch-ina”, and that it is an imitation of the one alreadypresent in the cathedral (Figure 7(c)).

According to Di Marzo, bas-reliefs of the two lunettes,top and bottom, which depict scenes from the healing ofthe paralytic (lower) and the miracle of Moses (above)were made by Ferraro.

Also inside the cathedral, in the chapel of the shrine, is

a work of gold with scenes in solid perspective: the silverark which contains the relics of St. Christina, createdin 1556 by sculptor goldsmith Paolo Gigli, assisted byFazio Gagini and Scipione Casella.

Finally, outside, leaning against the side walls of thesouthern portico are two additional solid perspective re-liefs. Both are bas-reliefs, the first commemorating thecoronation of Victor Amadeus of Savoy, King of Sicily,in 1713 made by G.B. Ragusa (1714), and the secondcommemorating the coronation of Charles of Bourbon(1735).

3 Methods of Acquisition andRestitution

In this first phase of work, innovative techniques forImage-Based Modeling (IBM) have been tested. Thephotogrammetric techniques, known since the late nine-teenth century, along with the advent of digital andtesting of new mathematical algorithms have achievedimpressive levels of processing to the point that today,thanks to research conducted by international groupsin the field of Computer Vision, it is possible to ex-trapolate three-dimensional models from photo shootsthrough the use of several IBM packages (Apollonio,Ballabeni & Gaiani, 2014; Bandiera, Beraldin & Gai-ani, 2011; Bertocci & Bini, 2012; Remondino & El-hakim, 2006; Remondino, Spera, Nocerino, Menna &Nex, 2014; Rodriguez Navarro, 2012; Toschi, Capra, De

10.7423/XJENZA.2015.1.10 www.xjenza.org

Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75. 72

(a) Baptismal Font, (b) southern font, (c) southern font,

Figure 7: Baptismal

Luca & Beraldin, 2014).Algorithms digitally translate photogrammetric tech-

niques, and lead to digital photogrammetry, retracingthe steps that lead to the stereoscopic model (interiororientation, relative orientation, absolute orientation).We have used the so-called algorithms of Structure fromMotion (SFM) that can reconstruct, from a sequenceof photographs, the internal parameters of the machineand the position in space of the homologous pointsthrough the recognition (manual or automatic) of cor-respondences between photographic images, generatinga set of three-dimensional points. While these SMFtechniques have previously been used for a purely tour-ist purpose on the web, they also have the potential togenerate great interest in reducing the time and costsin the acquisition and processing of data in the field ofcultural heritage.

Among the various types of IBM software, our at-tention has focused on two, both free and available fordownload, that work as web services: Autodesk 123DCatch (web application) and Recap (Autodesk 360).

The service acts in a semi-automatic way. The usercreates the project, chooses the resolution of the model(low, ultra), the smart cropping and/or texturing op-tion and the export format (eg OBJ). Images are thenuploaded and sent to the cloud. The user is notified byemail when the model is ready. They can then improve

the model themselves by adding survey points or refer-ence distance and manually stitching photos and thenresending it to the cloud. The model thus obtained, inOBJ format, can be imported into dedicated softwarefor 3D modeling and rendering.

A crucial phase is the realization of the photographicnetwork of the object: it is recommended that the anglebetween one shot and another is about 5-10 degrees sothat the overlap between neighboring frames is of about70 %; furthermore the images around the object shouldbe taken with different rotations and different heightsso as to vary the angle of the shoot.

The experiments conducted so far are very promisingeven when using amateur cameras. However, the use ofprofessional cameras with a fixed focal length and lowdistortion may improve the visual and metric quality ofthe obtained models.

4 Results and Discussion

Experiments carried out on various samples and on ar-chitectural scale, both from the building to the architec-tural detail, have demonstrated the visual and metric re-liability of these types of software, albeit with certain us-age limits and with the inescapable presence of the storeof knowledge of photogrammetry (Galizia & Santagati,2013; Casu & Pisu, 2013; Inzerillo & Santagati, 2013;Inzerillo, Santagati & Di Paola, 2013; Santagati & Inzer-

10.7423/XJENZA.2015.1.10 www.xjenza.org

73 Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75.

illo, 2013). In particular, for this object size the metricaccuracy has been tested in other studies by the authorsand is about 2-3/ millimeters and is a function of GDS(Ballabeni, Apollonio, Gaiani & Remondino, 2015).

Among the limitations, we stress the need to createa data set in which the photographic subject is framedin its entirety in a final usable model. However, if thesetypes of software on architectural scale show the limits ofa certain entity, on a small scale they have given qualityresults, allowing, in some cases (for example visualiza-tion purposes), the drop out of laser scanners, typicallyof high cost instrumentation and requiring specific ex-pertise and long processing times.

In this paper we report some applications carried outinside the Cathedral of Palermo (Figures 8, 9, 10 and11), that houses numerous sculptures of quadratura per-spective of great value.

Figure 8: Elaboration on the lower bezel “Healing of theparalytic” north basin.

Figure 9: Elaboration on the upper bezel “Blessing of theBaptismal Font” southern basin.

Figure 10: Elaboration on the memorial of the coronationof Charles of Bourbon.

Thanks to the use of these techniques, it will be pos-sible to return the perspective of these model works inquestion through a representation that allows the inter-pretation of the signs and construction (Figures 12 and

Figure 11: Elaboration on the upper bezel “Miracle ofMoses” north basin

13).

Figure 12: Processing one of the scenes of the ark of St.Christina

Figure 13: Restitution on the ark of St. Christina

5 ConclusionNowadays technology offers image based modeling pack-ages that allow novel approaches to the traditional 3Dacquisition and restitution of cultural heritage. The ob-tained 3D models via 123D Catch are reliable and accur-

10.7423/XJENZA.2015.1.10 www.xjenza.org

Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75. 74

Camera Resolution ofthe camera

Numberof images

Focal Distance(mm)

Calculationtime (minutes)

Healing of the Paralytic Nikon D 3200 12 Mpix 76 18 80Blessing of the baptismal font - - 20 35 30Commemorative Monument - - 25 24 37Miracle of Moses - - 27 18 40Ark of St. Christina - - 21 18 32

Table 1: Parameters pertinent the elaborations

ate both from the visual and the metric point of view.The benefits, in terms of lower costs and lower know-

ledge of photography and 3D modeling, are really sur-prising: even without having knowledge in the specificfield, you can get 3D models with the use of 123D Catch.

AcknowledgementsThe authors would like to thank Archivio Parrocchialeof the Cathedral of Palermo for having given us accessto the Cathedral’s premises.

References

Accascina, M. (1959). Aggiunte a Domenico Gagini.Roma: Istituto poligrafico dello Stato.

Amato, G. M. (1728). De principe templo panormitano.Palermo: Il Tempio dei Re.

Apollonio, F. I., Ballabeni, M. & Gaiani, M. (2014).Color enhanced pipelines for reality-based 3Dmodeling of on site medium sized archeological ar-tifacts. Virtual Archaeol. Rev. 5 (10), 59–76.

Ballabeni, A., Apollonio, F. I., Gaiani, M. & Re-mondino, F. (2015). Advances in image pre-processing to improve automated 3D reconstruc-tion. Int. Arch. Photogramm. Remote Sens. Spat.Inf. Sci. XL-5/W4, 315–323.

Bandiera, A., Beraldin, J. A. & Gaiani, M. (2011). Nas-cita e utilizzo delle tecniche digitali di 3D imaging,modellazione e visualizzazione per l’architetturae i beni culturali. In Ikhnos. ann. di anal. graf.e stor. della rappresent. (pp. 81–134). Siracusa:Lombardi.

Basile, N. (1926). La Cattedrale di Palermo : l’opera diFerdinando Fuga e la verita sulla distruzione dellaTribuna di Antonello Gagini. Florence: R. Bem-porad.

Basile, N. (1938). Palermo felicissima: divagazionid’arte e di storia. Palermo: Pietro Vittorietti Ed-itore.

Bellafiore, G. (1999). La Cattedrale di Palermo.Palermo: Flaccovio Editore.

Bertocci, S. & Bini, M. (2012). Manuale di rilievo ar-chitettonico e urbano. Novara: CittaStudi.

Casu, P. & Pisu, C. (2013). Photo-modeling and cloudcomputing. Applications in the survey of lategothic architectural elements. Int. Arch. Photo-gramm. Remote Sens. Spat. Inf. Sci. XL-5/W1,43–50.

D’Alessandro, M., Inzerillo, M. & Pizzurro, P. (1983).Omografia e prospettiva. Palermo: Istituto di Dis-egno e di Topografia della Facolta di Ingegneria diPalermo.

D’Alessandro, M. & Pizzurro, P. (1989). La Prospet-tiva solida nella scultura scenografica del Serpotta.Rilievi fotogrammetrici con restituzione grafico-numerica. (1st ed.). Palermo: Arti grafiche Giord-ano.

Di Marzo, G. (1880). I Gagini e la scultura in Sicilia neisecoli XV e XVI (1st ed.). Palermo: Tipografia delGiornale di Sicilia.

Di Marzo, G. (1884). I Gagini e la scultura in Sicilia neisecoli XV e XVI.

Galizia, M. & Santagati, C. (2013). Low-cost image-based modeling techniques for archaeological her-itage digitalization: more than just a good toolfor 3D visualization? Virtual Archaeol. Rev. 4 (9),123–129.

Giuffre, M. & Urbani, L. (Eds.). (1993). La Catted-rale di Palermo, studi per l’ottavo centenario dellafondazione. Sellerio Editore, Palermo.

Inzerillo, L. (2004). Procedure di costruzione della pros-pettiva nel progetto del disegno. Disegnare IdeeImmagin. 28, 84–91.

Inzerillo, L. (2012). Essere Prospettici. Be Perspective.Roma: Aracne.

Inzerillo, L. & Santagati, C. (2013). Il progetto delrilievo nell’utilizzo di tecniche di modellazionedense stereo matching. Disegnare Idee Immagin.47, 82–91.

Inzerillo, L., Santagati, C. & Di Paola, F. (2013).Image-based modeling techniques for architec-tural heritage 3D digitalization: limits and po-tentialities. In P. Grussenmeyer (Ed.), Xxiv int.cipa symp. (volume xl-5/w2) (pp. 555–560). Stras-bourg, France.

Kruft, H.-W. (1972). Domenico Gagini und seineWerkstatt. Munchen: Bruckmann.

10.7423/XJENZA.2015.1.10 www.xjenza.org

75 Inzerillo, L. and Santagati, C. (2015).Xjenza Online, 3:69–75.

Kruft, H.-W. (1980). Antonello Gagini und seine Sohne.Munchen: Bruckmann.

Mancino, I. (2007). Antonello Gagini fra Sicilia eMalta: il restauro delle statue della Cattedrale diPalermo. Caltanissetta: Fondazione culturale Sal-vatore Sciascia.

Remondino, F. & El-hakim, S. (2006). Image-based 3Dmodelling: A review. Photogramm. Rec. 21 (115),269–291.

Remondino, F., Spera, M. G., Nocerino, E., Menna, F.& Nex, F. (2014). State of the art in high densityimage matching. Photogramm. Rec. 29 (146), 144–166.

Rizzuti, R. (2007). La Tribuna di Antonello Gagini nellaCattedrale di Palermo. Palermo.

Rodriguez Navarro, P. (2012). La Fotogrametrıa DigitalAutomatizada frente a los sistemas basados en

sensores 3D activos. Rev. Expresion Grafica Ar-quit. 20, 100–111.

Santagati, C. & Inzerillo, L. (2013). 123D Catch: Effi-ciency, Accuracy, Constraints and Limitations inArchitectural Heritage Field. Int. J. Herit. Digit.Era, 2 (2), 263–290.

Toschi, I., Capra, A., De Luca, L. & Beraldin, J.-A.(2014). On the evaluation of photogrammetricmethods for dense 3d surface reconstruction in ametrological context. ISPRS Ann. Photogramm.Remote Sens. Spat. Inf. Sci. 2 (5), 371–378.

Vesco, M. (2011). Hic situs Emmanuel, plangite Sicelides. Le esequie reali di Emanuele Filiberto di Savoi-anella cattedrale di Palermo. Lexicon. Storie e Ar-chit. Sicil. e nel Mediterr. 13, 78–83.

Zanca, A. (1989). La Cattedrale di Palermo dalle originiallo stato attuale. Palermo: Accademia Nazionaledi Scienze Lettere e Arti di Palermo.

10.7423/XJENZA.2015.1.10 www.xjenza.org

Magri, D. C. (2015).Xjenza Online, 3:76–77.

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.orgDOI: DOI

News Article

2015 Researcher’s Mini-Symposium

D. C. MagriDepartment of Chemisty, Faculty of Science, University of Malta

Postgraduate researchers from the Faculties of Sci-ence, Engineering, Medicine & Surgery and Health Sci-ences gathered for a forum to present their research in-terests. The symposium was held in the afternoon of 30January 2015 in the Engineering Lecture Theatre.

The meeting consisted of two sequential sessions of13 speakers with an intermission for socialising and net-working between colleagues. Each researcher gave abrief 4 minute overview of their current research pro-jects. Some of the topics presented included piezoelec-tric materials, neural circuitry of the brain, structuresof proteins, colorimetric pH indicators, statistical mod-eling, apoptosis-inducing potential of curry powder ex-tracts on cancer cells, broadband antenna design, os-teoporosis, neurodegenerative research with fruit flies,aspirin-induced cell death, macrobenthic assemblages,genetic variation in the Maltese population, auxetic ma-terials, biogenesis in drosophila, design of micromirrors,co-crystallization of lanthanide complexes, surface coat-ing stainless steel, university students statistical viewon divorce, development of electrical circuits, rats beha-vior to nicotine, analysis of the human genome, surfacetreatment of biomedical titanium and molecular chap-erones.

The symposium promoted multi-disciplinary network-ing between various university faculties. Participantswere invited based on research topic diversity andgender balance. Refreshments during the intermis-sion were provided with financial support of the MaltaChamber of Scientists.

List of Postgraduate Researchers and Af-filiated Departments

L1: Rosalin Bonetta – Department of Physiology andBiochemistry

L2: Russell Farrugia – Department of Micro- and Nano-electronics

L3: Maria Cardona – Department of Chemistry

L4: Nowell Zammit – Department of Physiology andBiochemistry

L5: Lara Azzopardi – Department of Statistics and Op-erations Research

L6: Clayton Axiak – Department of Biology

L7: Melissa Formosa – Department of Applied Biomed-ical Science

L8: Eman Farhat – Department of Physics

L9: Michelle Briffa – Department of Physiology and Bio-chemistry

L10: Gianluca Farrugia – Department of Physiology andBiochemistry

L11: Lourdes Farrugia – Department of Physics

L12: Jeanesse Scerri – Department of Physiology andBiochemistry

L13: Marija Mangion – Department of Biology

L14: Joanna Vella – Department of Physiology and Bio-chemistry

L15: Keith Azzopardi – Metamaterials Unit

L16: Rebecca Borg – Department of Physiology and Bio-chemistry

L17: Jordan Gauci – Department of Micro- and Nano-electronics

L18: Stephanie Ghio – Department of Physiology andBiochemistry

L19: Duncan Micallef – Department of Chemistry

L20: Sarah Farrugia – Department of Metallurgy & Ma-terials Engineering

L21: Lara Montebello – Department of Statistics andOperations Research

Correspondence to: D. C. Magri (david.magri@um.edu.mt)

c© 2015 Xjenza Online

77 Magri, D. C. (2015).Xjenza Online, 3:76–77.

Figure 1: The poster is compiled from figures from the 26 presentations given by postgraduate researchers at the Researcher’sMini-Symposium on the 30th January 2015 held in the Engineering Lecture Theatre at the Msida campus, University of Malta.

L22: Barnaby Portelli – Department of Micro- andNanoelectronics

L23: Caitlin Davies – Department of Physiology andBiochemistry

L24: Clint Mizzi – Department of Physiology and Bio-chemistry

L25: Michelle Cortis – Departments of Anatomy andMetallurgy and Materials Engineering

L26: Marita Vella – Department of Physiology and Bio-chemistry

DOI www.xjenza.org

Xjenza Online - Journal of The Malta Chamber of Scientistswww.xjenza.org

News Article

The New Malta Neuroscience Network Programme

Interview with the Coordinator Prof. Giuseppe Di GiovanniUniversity of Malta, Department of Physiology and Biochemistry

“The human brain has been described as the “lastfrontier” in medical research and one of the most diffi-cult areas in medicine. Unravelling the brain’s secretscould change the lives of millions of people of all agessuffering from neurological and psychological conditions,addictions and lesions of the central nervous system forthe better. Brain diseases can affect anyone of any age,socio-economic status or gender. One in three Maltesepeople and about 1 billion people worldwide will findthemselves having to face a psychiatric or neurologicaldisease at some point in their lives, and while treatmentsexist, currently there are no cures. From autism to mul-tiple sclerosis and Alzheimer’s disease to brain trauma,brain disorders represent the most important challengeto public health in the 21st century” says Professor DiGiovanni, Co-ordinator of Malta Neuroscience Network.

“With the creation of the Malta Neuroscience Net-work, we will be keeping up with the most import-ant developments with regard to brain research world-wide: multi-disciplinary collaboration. Understandingthe way the brain works, and above all brain diseases isextremely complicated, and requires the involvement ofresearchers coming from a number of different scientificdisciplines and clinical areas collaborating in new ways.Just as important is the participation of patients, famil-ies and health workers as well as the organisations thatrepresent them such as the Richmond Foundation andwe hope many more” continues Professor Di Giovanni.

Currently, we are in a time which scientists havedefined “The Golden Age of Neuroscience”. Revealingthe brain’s secrets is a global concern involving scientistsworldwide. Now we have a much deeper understandingof the complexity of the brain which has greatly im-proved human health. Over the last few years, we havemade important discoveries, like for example, better-ing our understanding of the mechanisms that underliehuman consciousness, we have discovered mirror neur-ons and created technologies which allow communica-

tion between the brain and computers (brain-computerinterface). In addition to these innovative developments,we have increased of understanding of the genetic basisof complex diseases such as autism, schizophrenia, Par-kinson’s disease and Alzheimer’s.

While these efforts are promising, it is important torealise that success will be most easily assured if re-searchers receive funding which allows for them trans-late their research into a clinical setting enabling themto work with clinical researchers ensuring that new dis-coveries pass from the bench to the bedside. This is theonly approach which will get us closer to our object-ives of understanding the brain and its incredible influ-ence more fully and which will allow us to protect brainhealth and therefore benefit patients, their families andhealth workers.

The challenges facing research in neuroscience todaycan only really be overcome with strong support fromthe community (government and society). Global col-laboration efforts such as the BRAIN initiative in theUSA and the Human Brain Project in Europe havebeen fundamental in fast tracking discoveries in brainresearch. Malta is a key player and in fact many Malteseresearchers publish high quality research on the brain.This has happened despite the Maltese Government’s,relatively small investment in neuroscience research inthe last ten years.

There has also been wide debate with regards tomaximising the impact that research funding can have:should the focus be on basic research or research whichis in an advanced stage with commercial potential? Theanswer is logical: both. However, the Government hasdecided to finance the applied one exclusively. In truth,research is a process and progress is necessary at everystage to enable new or better diagnoses, preventionstrategies, and treatments and finally cures. We needto understand how the brain works as a single systemfrom the biological bases of the disease, with the aim

c© 2015 Xjenza Online

79

of identifying and developing potential diagnostic toolsand therapies.

“With the creation of Malta Neuroscience Networkand its Research Fund by RIDT will raise awareness andunderstanding of the brain and brain disease, and in-crease support for Malta’s excellent and innovative brainresearch,” continued Prof. Di Giovanni. “Every Maltesewill benefit from Malta’s leadership and vision in thisimportant area. We hope that the Government of Maltawill match on 1:1 basis, private and non-governmentaldonations. The Fund will be first single investment inbrain research ever made in Malta.

There has never been a more exciting time to be partof the global brain research and brain health community.We thank all of our supporters over the past year—andespecially our dedicated Board and staff. We look for-ward to advancing Malta’s central role in the comingyear.”

Information about the MaltaNeuroscience Network“The new Malta Neuroscience Network is a web basedcommunity made up of researchers (neurologists, psy-chiatrists, radiologists, biomedical scientists, engineers,psychologists, cognitive and ICT scientists. . . ) and oth-ers from different faculties at the University of Maltaworking on neuroscience. Prof Giacomo Rizzolatti hasagreed to be present at the launch of the Brain Aware-ness Week (December 1-6) and will be attending theMalta Medical School Conference (IX MMSC). He isone of the most important scientists in the world; heis then MD, neurologist who discovered mirror neurons.His discovery has changed Neuroscience in the same waythat the discovery of DNA changed biology.” He will beawarded by the Rector the Affiliate Professorship andhe will be list among the members of our Network.

The Programme for Malta Neuroscience Network, un-der the aegis of the Faculty of Medicine and Surgery, is ajoint initiative among the Faculty of Medicine and Sur-gery, Faculty of Engineering, Faculty of Health Sciences,Faculty of Information & Communication Technology,Faculty of Media & Knowledge Sciences, Faculty of Sci-ence, the Faculty for Social Wellbeing and the Centrefor Biomedical Cybernetics and the Centre for Molecu-lar Medicine and Biobanking.

1. The aims of the Program are:

(a) To encourage and facilitate interdisciplinaryresearch that brings together academic mem-bers from all the Faculties of the University ofMalta with an interest in in the rapidly grow-ing field of Neuroscience.

(b) To promote interdisciplinary dialogue amongall the disciplines involved with Neuroscience.

(c) To foster research and training in neuroscienceat University of Malta.

(d) To hold regular meetings, seminars and con-ferences (Neuroscience day @ University ofMalta) in which to present research ideas, dis-cuss work in progress and generally promotethe sharing and dissemination knowledge.

(e) To sponsor and coordinate seminars by leadingneuroscientists from home and abroad.

(f) To create and maintain an electronic portal forthe publications, discussion and disseminationof research.

(g) To offer study-units in Neuroscience that maybe included in both undergraduate and post-graduate programmes.

(h) To collaborate with local and overseascentres/Universities, programmes and indi-viduals with similar purpose and scope.

(i) To raise public awareness in Neuroscience,brain disorders and mental health and otherrelated areas through public talks, eveningcourses, Annual Brain Awareness Week, BrainResearch Fundraising in conjunction withRIDT and different NGOs and scientific as-sociations.

2. The Programme has the following structure:

(a) A coordinator appointed by Council (Prof Gi-useppe Di Giovanni) and a Board appoin-ted by Senate on the recommendation of theFaculty Boards of Faculty of Medicine andSurgery, Faculty of Engineering, Faculty ofHealth Sciences, Faculty of Information &Communication Technology, Faculty of Me-dia & Knowledge Sciences, Faculty of Science,the Faculty for Social Wellbeing, the Board ofthe Centre for Biomedical Cybernetics and theCentre for Molecular Medicine and Biobank-ing.

(b) The Board consists of the Coordinator andeighteen other members, two each from theFaculty Medicine and Surgery, Faculty of En-gineering, Faculty of Health Sciences, Facultyof Information & Communication Technology,Faculty of Media & Knowledge Sciences, Fac-ulty of Science and the Faculty for Social Well-being and from the Centre for Biomedical Cy-bernetics. Chairpersonship of the board shallbe held jointly, with the joint chairpersons be-ing chosen by and from among the members.

(c) The appointment of the Coordinator and themembers of the Board is for one year, whichcan be renewed.

3. The MNN Programme will operate as a cost

www.xjenza.org

80

centre with the ability to raise its own funds, sub-ject to the University’s financial regulations. Anedited extract from this interview with Prof Di Gio-

vanni is published in July issue of the magazineThink.

Malta Neuroscience Network Programme (MNN) GOVERNACE

COORDINATOR:

PROFESSOR GIUSEPPE DI GIOVANNI,Faculty of Medicine and Surgery, Department of Physiology and Biochemistry.

BOARD

FACULTY OF MEDICINE AND SURGERYDr Norbert R VellaDr David Cassar

FACULTY OF ENGINEERINGDr Tracey CamilleriProf Simon Fabri

FACULTY OF HEALTH SCIENCESProf Angela A. Xuereb AnastasiProf Helen Grech

FACULTY OF INFORMATION & COMMUNICATION TECHNOLOGYDr George AzzopardiDr Lalit Garg

FACULTY OF MEDIA & KNOWLEDGE SCIENCESDr Noellie BrockdorffProf Ian Thornton

FACULTY OF SCIENCEProf Charles SammutProf Camilleri Liberato

FACULTY FOR SOCIAL WELLBEINGProf Carmel CefaiDr Kristina Vella

CENTRE FOR BIOMEDICAL CYBERNETICSProf Kenneth CamilleriDr Owen Falzon

CENTRE FOR MOLECULAR MEDICINE AND BIOBANKINGProf Richard MuscatProf Mario Valentino

www.xjenza.org

XJENZAVOLUME 3 ISSUE 1, 2015

ONLINE

Table of Contents

Articles

1 Malta’s participation in Europe’s oldest research cooperation frame-work

J. Mifsud

Special Issue Introductory Article6 ‘Everyone is a winner, help is just a push

of a button away . . . ’:The Telecare Plus service in Malta

Research ArticleM. Formosa

17 Connecting Frontier Research with Industrial Development - Lanthanide and Actinide Chemistry in the European f-Element Network

Review ArticleU. Baisch

23 Real-Time Modelling and Interpolation of Spatio-Temporal Marine Pollution Research ArticleN. Formosa and K. Scerri

31 Towards Sound Refactoring in Erlang Research ArticleE. Tanti and A. Francalanza

36 Symbiotics of History and Social Psychology Understanding Social Representations of History in Europe Research ArticleS. Azzopardi, E. Buttigieg and L. Licata

41 Numerical Modelling of Soil Erosion Susceptibility in the Maltese Islands using Geographic Information Systems and the Revised Universal Soil Loss Equation (RUSLE)

Research ArticleD. Sultana

51 Typing Actors using Behavioural Types Research ArticleJ. Masini and A. Francalanza

56 Assessing potential output growth of the maltese economy using a production function approach

Research ArticleA. George Grech and B. Micallef

63 Digital Technologies for virtual recomposition. The case study of Serpotta stuccoes

Research NoteF. Di Paola

69 Architectural Perspectives in the Cathedral of Palermo: Image-Based Modeling for Cultural Heritage Understanding and Enhancement Review Article

L. Inzerillo and C. Santagati

76 2015 Researcher’s Mini-Symposium News ArticleD. C. Magri

78 The New Malta Neuroscience Network Programme

News ArticleG. Di Giovanni