responsive open learning environmentsrole-project.archiv.zsi.at/wp-content/uploads-role/... ·...

Integrated Project

Responsive Open

Learning Environments European Commission Seventh Framework Project (IST-231396)

Deliverable D7.2

Strategies and Facilities for Activity Pattern Sharing

Editor Felix Mödritscher, Fridolin Wild, Zinayida Petrushyna

Work Package WP7

Status Final

Date February 19, 2010

of 50

of 50

The ROLE Consortium

Beneficiary Number

Beneficiary name Beneficiary short name Country

1 Fraunhofer FIT FHG DE

2 RWTH Aachen University RWTH Aachen DE

3 Technical University of Graz TUG AT

4 Katholieke Universiteit Leuven K.U.LEUVEN BE

5 University of Koblenz UNI KO-LD DE

6 Uppsala University UU SE

7 École Polytechnique Fédérale de Lausanne

EPFL CH

8 University of Leicester ULEIC UK

9 Open University UK OU UK

10 Vienna University of Economics & Business

WU AT

11 Festo Lernzentrum Saar GmbH FESTO DE

12 imc AG IMC DE

13 British Institute for Learning and Development

BILD UK

14 Shanghai Jiao Tong University, China

SJTU RPC

15 Zentrum für Soziale Innovation ZSI AT

16 U&I Learning UIL BE

of 50

Document Control

Title: Strategies and Facilities for Activity Pattern Sharing

Editors: Felix Mödritscher (WU), Fridolin Wild (OU), Zinayida Petrushyna (RWTH Aachen)

E-mail: [email protected], [email protected], [email protected]

Amendment History

Version Date Author/Editor Description/Comments

1 Dec 7, 2009 Felix Mödritscher First draft (structure, introduction)

2 Dec 17, 2009 Felix Mödritscher Section 2 (PLE competences), Section 3 (patterns)

3 Jan 13, 2010 Fridolin Wild, Felix Mödritscher

Revision of structure, Section 1 (introduction), Section 3 (pattern capturing and sharing)

4 Jan 21, 2010 Felix Mödritscher Section 3 of revised document

5 Jan 26, 2010 Felix Mödritscher, Carsten Ullrich

Improvements by SJTU on Section 3, first parts of Section 4 (pattern repository)

6 Jan 28, 2010 Felix Mödritscher Final version of Section 4, Section 5 (summary)

7 Feb 2, 2010 Felix Mödritscher, Zinayida Petrushyna

Preparation of version for the internal review, input and improvements by RWTH

8 Feb 19, 2010 Felix Mödritscher, Fridolin Wild, Zinayida Petrushyna

Consideration of ROLE reviewers, finalisation of the deliverable

Contributors

Name Institution

Zinayida Petrushyna RWTH Aachen

Sandy El Helou EPFL

Fridolin Wild, Alexander Mikroyannidis OU

Felix Mödritscher WU

Carsten Ullrich SJTU

Joanna Wild, Margit Hofer CSI

Marco Kalz, Marcus Specht OUNL

mailto:[email protected]



of 50

Reviewers

Name Institution

Effie Law ULEIC

Erik Duval K.U.Leuven

Legal Notices

The information in this document is subject to change without notice.

The Members of the ROLE Consortium make no warranty of any kind with regard to this document, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. The Members of the ROLE Consortium shall not be held liable for errors contained herein or direct, indirect, special, incidental or consequential damages in connection with the furnishing, performance, or use of this material.

of 50

Executive Summary Setting a personal learning environment (PLE) approach into practice is not trivial, as the proposed end-users have varying attitudes towards and experiences in using this kind of technology. In fact, applying different learning tools provided by educational organisations, by companies or through the Internet requires that lifelong learners develop not only professional competences but also hands-on skills as well as personal and social competences, the so-called transcompetences. Thus, a PLE solution has to provide scaffolding and support facilities to empower all kinds of learners to design and use their environments so that they can connect to learner networks, collaboratively work on shared artefacts, and successfully achieve their goals.

This deliverable aims at examining strategies and facilities for supporting learners in using PLEs by providing them ‘good practices’, i.e. learning experiences created by peer actors. Specifically for PLE-based learning activities and due to trust and privacy considerations, these practices can be captured in the form of activity patterns distilled and abstracted from learner interaction recordings by removing sensitive data. Thereby, pattern-based good practice sharing is approached from three perspectives. First of all, PLE-related competences are identified through an empirical study. Consequently, strategies and facilities for capturing and sharing activity patterns are elaborated based on state-of-the-art technology and approaches. Finally, the document gives an overview of existing technology applicable as a pattern repository.

Findings of the user study and the technology surveys can be summarised as follows:

• Working with PLEs requires professional and transcompetences, whereby activity patterns contain all these competences implicitly. Furthermore, the development of competences is triggered by interactions with the learning environment, e.g. given through activity patterns, and can be fostered through information pull and push strategies and through facilities aiming to support specific transcompetences identified by empirical studies. With respect to experiences from the field of epistemic games, good practice sharing through activity patterns seems to be useful for inexperienced PLE learners.

• Approaches for activity pattern capturing and sharing are already available, although these strategies and techniques come from other disciplines or contexts. None of the reviewed solutions fulfils all requirements for pattern sharing, whereby particularly the concept of de-personalization (i.e. anonymizing and removing other sensitive data from interaction recordings) is hardly realised in practice. Moreover, pattern capturing and sharing can be characterised along six dimensions, namely architectural design, activity structure, interaction type, tracked applications, social form, and de-personalization.

• A technology survey shows that practice sharing infrastructures (including pattern repositories) have been implemented and experienced in other application areas. Consequently, this deliverable proposes to enrich the PLE infrastructure by pattern repositories which can be plugged to the existing PLE solutions and have a specialised API for publishing and retrieving good practices in the form of activity patterns. Such pattern repositories should provide facilities for manual

of 50

sharing of pre-configured PLEs (the patterns) and automated strategies, like recommending interactions or tools.

• The usefulness, the efficiency, and the effectiveness of a pattern sharing infrastructure for PLEs will be evaluated if adequate technology is available.

This deliverable relates to the first four WP objectives (cf. ROLE-DoW, p. 58), whereby the main focus lies on the second and the fourth objectives:

• Developing a methodology for building and altering networks of actors, artefacts, and activities on the basis of PLEs

• Empowering learners with learning environment design capabilities and facilitating learnability for creating PLE mash-ups

• Providing regulation and reflection facilities for collaboration in learning networks

• Facilitating best practice sharing

It is an outcome of the tasks 7.1 and 7.2 (cf. ROLE-DoW, p. 59) and intended for pedagogical experts and technologists in the field of technology-enhanced learning. Within the ROLE project, it attempts to fulfil the objectives RO1, RO2, and RO3 (cf. ROLE-DoW, p. 6):

• RO1: support the individual assembly of accessible learning services, tools and resources in responsive open learning environments (ROLE)

• RO2: research and develop a psycho-pedagogically sound framework for supporting the individual composition of learning services in ROLE

• RO3: create new engineering methodologies to enable significant contributions to ROLE from learner and developer communities from outside the project consortium

This document relates to the deliverables D7.1/ID7.2 (“Model and Methodology for PLE-Based Collaboration in Learning Ecologies”) and ID7.1 (“Draft Prototypes of a Mash-up PLE”). The key features identified in this document are considered as input to the deliverable D1.5 (“Revised requirements specification”). Parts of section 2 were published at the MUPPLE workshop of the European Conference on Technology Enhanced Learning (EC-TEL 2009) conference (see Wild et al., 2009).

of 50

Table of Contents

EXECUTIVE SUMMARY 5

1 INTRODUCTION 8

1.1 Good practice sharing through activity patterns 8

1.2 Example user scenario for pattern sharing 11

1.3 Methodological approach and structure of the deliverable 12

2 THE COMPETENCE CONTINUUM FOR PERSONAL LEARNING ENVIRONMENTS 13

2.1 State-of-the-art overview 13

2.2 A case study on PLE-related competences 14

2.3 Summary on strategic competences in connection with PLEs 17

3 CAPTURING AND SHARING ACTIVITY PATTERNS 19


3.2 Dimensions of pattern capturing and sharing 21

3.3 Summary on state-of-the-art strategies and facilities 24

4 PATTERN REPOSITORIES 26


4.2 Related work and discussion 29

4.3 Summary on repository strategies and facilities 32

5 SUMMARY ON POSSIBLE STRATEGIES AND FACILITIES 33

5.1 Conclusions of the case study and the technology surveys 33

5.2 Proposed strategy and facilities for sharing good PLE practices 34

5.3 Applicability in PLE solutions and further action plan 35

APPENDIX A: SURVEYS OF PATTERN SHARING TECHNOLOGY 36

A.1 Survey of technology for activity pattern sharing 36

A.2 Survey of technology for activity pattern capturing 39

A.3 Key features for activity pattern sharing 43

REFERENCES 47

of 50

1 Introduction According to Henri et al. (2008), personal learning environments (PLEs) refer to a set of learning tools, services, and artefacts gathered from various contexts to be used by the learner who designed the environment. However, user studies in the field of higher education (Nguyen-Ngoc & Law, 2008) and workplace learning (Kooken et al., 2007) evidence that learners – and even teachers (Windschitl & Sahl, 2002) – have varying attitudes towards and hand-on skills in using ICT for learning. On the one hand, they may be capable of adopting and utilising new tools for their purposes and needs easily. On the other hand, ICT may restrict them as they spend too much time on playing with the tools, being unfocussed when using them, or even failing to achieve their goals due to frustration and distraction trying to handle them (Windschitl & Sahl, 2002).

Such negative feelings hinder learners from proceeding with their learning as they cannot adapt their environments according to their needs and goals. Simply said, ICT can be a distracter for learning and the learning progress (see also Ben Youssef & Dahmani, 2008). As stated by Van Harmelen (2006, 2008), personal learning environments aim at empowering learners to design (ICT-based) environments for their learning activities and acquire competences through using the PLE and not being frustrated by ICT usage.

Eckstein et al. (2001) outline the necessity of capturing and sharing successful teaching practices, i.e. through pedagogical patterns, so that instructors can set didactical strategies and translate them into practice without going through the time-consuming process of consulting didactical experts. Similar findings on pedagogical patterns are reported for CSCL processes (Persico et al., 2009) and learning efficiency (Kolfschoten et al., 2010). In accordance with these experiences, practice sharing seems to be a critical requirement for personal learning environment (PLE) settings, as it can ease ICT usage and reduce frustration from working with technology. Furthermore, PLE-related practices do not only focus on developing professional competences but also on fostering so-called transcompetences which comprise hand-on skills for learning tools, self-regulated learning skills or social competences (cf. Henri et al., 2008; Wild, 2009).

1.1 Good practice sharing through activity patterns In the scope of this document, practice sharing is examined under the premises of learning with new technologies and in networked communities. Therefore, practices can be characterised by two special aspects. Firstly, a practice in connection with a PLE-based, collaborative activity requires all kinds of competence, ranging from domain-specific knowledge through hand-on skills for the tools to social and personal skills. Secondly, practices are primarily provided by peer actors independently of their role (learner, teacher, facilitator, knowledge worker) and not necessarily by experts. Thus, this deliverable does not refer to ‘best practices’ but only to ‘good practices’ created by end-users. This kind of practice can be useful for one actor and inappropriate for another. It is highly dependent on a learner’s characteristics, needs, present competences and surely on the usage context.

of 50

Above and beyond, this deliverable aims at examining strategies and facilities for supporting PLE users by providing experiences (good practices) from those who are capable of successfully handling ICT tools for their lifelong learning activities and willing to share. The core idea for achieving good practice sharing is that PLE usage experiences can be recorded, shared, and re-used by others again. The practice sharing considerations are based on the premises that the PLE competences are latent components of lifelong learning activities which have been researched for aspects like soft and team competences (Motschnig-Pitrik & Figl, 2008), self-directed learning capabilities (Law & Nguyen-Ngoc, 2008), or ICT hands-on skills (Windschitl & Sahl, 2002). Moreover, interaction sequences of successful PLE-based activities could be a source to study transcompetences and professional ones.

The original recordings of learner interactions, as elaborated in the ROLE Deliverable D7.1/ID7.2 (“Model and Methodology for PLE-Based Collaboration in Learning Ecologies”), should not be shared due to two important reasons, namely trust and privacy considerations. Trust can be understood as the “the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party” (Mayer et al., 1995). Thus, it is recommended to give the learners full control over this sensitive data. Sharing these recordings should be initiated or, at least, be permitted by the learner.

Apart from that, privacy is defined as “the interest that individuals have in sustaining a ‘personal space’, free from interferences by other people and organizations” (Clarke, 2006). Digital recordings of learner interactions are part of this personal space and should be secured to preserve the learners’ privacy, which particularly is necessary for open(-content) systems (cf. García-Barrios, 2009). Hereby, the approach for capturing good practices starts with capturing learner interactions and continues with distilling and abstracting them into a so-called ‘activity pattern’ if the user considers the activity to be a successful one which might be helpful for others as well.

According to Alexander et al. (1977), a pattern “describes a problem which occurs over and over again in our environment, and then describes the core of the solution to that problem, in such a way that you can use this solution a million times over, without ever doing it the same way twice”. Similarly, activity patterns can be understood as “archetypal and reusable recordings of design decisions taken by the users or developers who created a learning environment” (adopted from Alexander et al., 1977), i.e. a recording of a learner interaction sequence with (partially) removed entities.

Activity patterns can reach from a single learner interaction, e.g. the selection or visual arrangement of a tool, up to a pre-configured PLE for a specific situation (an activity) which even can involve several actors. Additionally, Sobernig et al. (2006) outline that “design activity often is a construction process that aims at building constructs and conceptual models from [learning] experience”. Figure 1-1 visualises the process of good practices in PLE settings which is supposed to be a lifecycle. On the left side of the figure, users who are experts in their domain and have the required competences and skills utilise the learning tools available to achieve their goals given by lifelong learning activities like learning-on-the-job or further education.

Fig. 1-1: Good practice lifecycle in PLE settings

The first important process in the good practice lifecycle deals with recording interactions of learners with their environments and has been elaborated in the ROLE Deliverable D7.1/ID7.2. Digitalisation of PLE interactions enables learners to share their experiences. According to Dillenbourg (2005), such interaction recordings contain rich information on e.g. the context of messages, knowledge sharing, sharing of construction of understanding, etc., which can be useful for various other purposes like analysing methods or good practice sharing. As mentioned before, parts of the recordings might be sensitive or personal and, therefore, should be secured and controlled by the end-users. In the context of this document, the process of anonymizing learner interactions is called de-personalization and leads to digitalised learning experiences with removed or masqueraded parts, the activity patterns.

Consequently, the PLEs and interactions of experienced PLE users are available to peers, either through own repositories, manual disclosure of the activity patterns, or automated approaches like recommendations mined from the patterns. Now, it should be possible that other PLE users can find and re-use (re-personalize) these patterns. By re-using good practices of peers, the lifecycle starts with the recording of user interactions again, leading to a new version of the pattern which can be slightly modified or completely different from the old one.

Figure 1-1 indicates that good practice sharing does not only focus on professional competences but also on transcompetences, as hands-on skills and soft competences are implicitly contained in the patterns. Moreover, they are situated and context-bound, each pattern standing for an activity experienced by one or more actors in a specific situation. Even patterns of ineffective practices might be of interest to learn about failures and mistakes. Finally, the activity patterns available can be analysed according to networked collaboration or recommendations of pre-defined PLEs and single user interactions.

of 50

1.2 Example user scenario for pattern sharing A typical situation for PLE-based collaboration is depicted in Figure 1-2. A learner is involved into two activities, an individual tutoring session in which she consults the facilitator via Skype and a task in which she collaboratively creates an artefact together with a peer actor. Good practice sharing starts after one of these activities has been completed.

Fig. 1-2: Example user scenario for PLE-based collaboration, taken from Wild (200)

For instance, a learner (the ‘peer’ actor) contributes to a joint paper by searching literature using the ObjectSpot tool, reading the literature using various reader tools (e.g. Adobe Acrobat Reader or Microsoft Word), communicating with the other authors via Webmail, and submitting her input to a Wiki page in the XoWiki system. After having (successfully) finished these interactions, she is satisfied with this experience and wants to share it without exhibiting sensitive data and without spending too much effort.

So, she labels the pattern with the title ‘contributing to a conference paper’ and submits it to a pattern repository of her choice. The repository warns the learner that there is sensitive data, i.e. the link to the joint paper, in the recording. Thus, she replaces this link with a placeholder which requires pattern consumers to fill out the link to their own paper. After de-personalizing this pattern, she successfully published it on the selected pattern repository.

A few days later, another learner is invited to join a collaborative paper writing activity. As this learner has no experiences how collaboration with the new colleagues looks like and which tools they use, he simply asks them. One of them consults the pattern repository and sends him a reference to the activity pattern mentioned above. After loading the pattern into his PLE client, he has to specify the URL of the paper to write. Furthermore, he decides to remove the Webmail system and use his own email client. Finally, he tags ObjectSpot as an important search engine for the context of writing a paper with these colleagues.

of 50

of 50

When entering his paper contribution into the shared Wiki page, the PLE client receives two recommendations of single learner interactions from the pattern repository, namely a dictionary service and a link to the glossary of the research colleagues. The learner decides to add the glossary to his environment in order to avoid terminological inconsistencies in his paper contribution. Having finished this activity, he also decides to share this experience as a pattern.

1.3 Methodological approach and structure of the deliverable With respect to the good practice lifecycle visualised in Figure 1-1, this deliverable examines the core processes of and technology for activity pattern sharing and proposes strategies and facilities for a PLE infrastructure. Therefore, each section gives a theoretical overview of the pattern sharing issue. Then, the state-of-the-art for the section is evaluated either on the basis of an empirical study or through technology surveys. Finally, each section outlines strategies and facilities for practice sharing in open learning environments. The rest of the document is structured as follows:

Section 2 elaborates the competences required for successfully using PLE technology. Backed by an empirical user study, this section addresses the competences beyond the professional ones, i.e. the transcompetences, which comprise hands-on skills for using PLE technology as well as social and personal competences.

In section 3 the most relevant aspects of capturing and sharing activity patterns are summarised, and state-of-the-art technology is reviewed. As a result, this section leads to dimensions relevant for capturing and sharing patterns in PLE settings, before possible strategies and facilities are highlighted.

Then, section 4 examines existing technology which can be utilised as a pattern repository. Besides discussing the relevant aspects of pattern repositories, approaches and key features are derived for an approach within the ROLE project.

Finally, section 5 concludes the deliverable by sketching a holistic approach towards activity pattern sharing in PLE settings. Hereby, a possible architecture enabling good practice sharing is proposed, and the next steps are outlined.

of 50

2 The Competence Continuum for Personal Learning Environments

In the beginning personal learning environments (PLE) were strongly motivated by their opposition to learning management systems, while today they form a movement on their own. PLEs provide a perspective on learning environments that focuses on the individual (not the institution): they envision an empowered learner aiming for self-direction for whom tightly- and loosely-coupled tools facilitate the process of defining outcomes, planning and assessing their achievement, constructing knowledge, and regulating learning (Van Harmelen, 2008) – either collaboratively or independently.

Because of the gist of PLEs to be individual software or a bunch of software tools, the observation of single person practices should be easy to implement. Thus, activity pattern capturing and sharing is dependent on the possibilities of personalizing learning processes. Furthermore, according to Henri et al. (2008), resources used in formal education to support metacognition, self-direction, and reflexivity should be reconceptualised and redesigned in PLE tools that may play a key role in competence acquisition of learners in the near future.

2.1 State-of-the-art overview While there is currently an intense discussion about the concept of personal learning environments and its technological and organisational foundations, it remains an open question whether a particular set of skills or competences is required for facilitators and learners to use PLEs in their education or educational design activity. This question is strongly related to the 'media literacy' and ‘digital literacy’ discussion in educational technology. Thoman & Jolls (2005) define media literacy as a “21st century approach to education. It provides a framework to access, analyse, evaluate, and create messages in a variety of forms – from print to video to the Internet. Media literacy builds an understanding of the role of media in society as well as essential skills of inquiry and self-expression necessary for citizens of a democracy.”

In their declaration on lifelong learning in 2006 the European Commission mentions eight key competences. One of these competences is defined as a “digital competence [that] involves the confident and critical use of information society technology (IST) and thus basic skills in information and communication technology (ICT)” (European Parliament and the Council of Europe, 2006). Furthermore, a digitally literate person is “equipped with the skills to benefit from and participate in the Information Society. This includes both the ability to use new ICT tools and the media literacy skills to handle the flood of images, text and audiovisual content that constantly pour across the global networks” (EC, 2007). Learners with a high level of digital literacy deploy ICT efficiently depending on situation and aim, use them to generate information and knowledge in their profession, and transform knowledge and practice through innovation and creativity with the help of these technologies (DTI, 2007).

of 50

The eSkills model by Stucky et al. (2003) determines five categories of skills: (1) IT awareness (basic knowledge), (2) IT literacy (knowledge to operate a PC), (3) expert user (special knowledge or expertise with application software, helping other users), (4) professional entry level (professional knowledge), and (5) professional level IT skills (advanced professional knowledge). Different than media and digital literacy, eSkills primarily focus on instrumental aspects and provide a model for developing proficiency in applied settings.

With the rise of social software in the last years and technologies for a programmable web, a silent revolution has taken place that promotes a new era of end-user friendliness and usability that should enable users to access, manage, re-mix, create, and communicate information and knowledge to various individuals and networks more efficiently. At the same time, the usage of Internet technology, like email, blogs, wikis, social networking platforms, etc., fosters public exposure and the ever-growing collection of digital traces of users. This data distributed throughout the Web can be a source of valuable information for constant improvement of the user’s eSkills, e.g. by providing support facilities for learning, like intelligent feedback, reflection, recommendations or predictions. Examples are given with community approaches like Mendeley or RescueTime in the next section.

Overall, media literacy is the only one of the competence clusters that can be acquired and evolved by PLE users. The other types of competences have to be considered in the continuum of PLE usage, like self-directed learning, competences to plan, organise and manage educational activities, meta-cognitive competences, competences to evaluate the quality of one’s knowledge and leaning strategies, etc. (see. also Henri et al., 2008). It is still, however, an open question, which skills are needed to apply these technologies in an educational context.

Considering PLE-related competences that are acquired during PLE usage and good practice sharing, Shaffer et al. (2009) examine how to measure the evolvement of learners within digital learning systems and the usefulness of practice sharing. For example, in the Digital Zoo game for the secondary school level, Shaffer and his colleagues collected skills, knowledge, values, identities, and epistemology (awards, roles or positions within the learning community) of pupils and mentors. After several hours of practicing the same data was gathered again. Shaffer et al. applied epistemic network analysis (ENA) to identify positions of different competence types in epistemic frame that may be considered as ‘competence-map’. Thus, it is possible to sketch the differences between the competence-maps over the course of the game. The most interesting fact was that the pupils began to imitate their mentors by generating similar competence-maps. This indicates the usefulness of good practices which the Digital Zoo mentors passed to pupils within the scope of epistemic games (formal education) but which may be conceptualised in PLE setting as well.

2.2 A case study on PLE-related competences Wild et al. (2009) describe an empirical study with European instructors attending an international training workshop on the application of social software for learning. The study aims at shedding light on the competences (skills, abilities, attitudes, habits,

knowledge) required and motivated by PLE approaches. Presumably, the findings of the study provide some valuable input on how literacies and skills are developed with the help of these environments.

As shown in Figure 2-1, five dimensions with three stages each could be identified within the user interviews conducted. These dimensions encompass: plan, reflect, monitor, act, and interact. The stages distinguish through the interviews partition the underlying competences (skills, abilities, attitudes, habits, knowledge) into those that serve as a minimal condition, necessary triggers, and intended outcomes.

Fig. 2-1: The PLE competence continuum

of 50

of 50

The term 'minimal condition' refers to those skills, abilities, attitudes, habits, and knowledge that need to be present or are typically present when beginning to consciously build a PLE, whereas 'necessary triggers' denote those that are developed along the way towards the 'intended outcomes' and on which the intended outcomes rely. They can be surrogate or just incomplete in nature.

In brief, planning competence refers to those skills, abilities, habits, attitudes, and knowledge that fix how goals, schedules, and paths are set. Reflection is constructive sense making of the past and enables planning. Monitoring refers to how progress control is performed. Last but not least, the pair acting and interacting group ‘social & collaboration’ as well as ‘information & tool’ competences.

Discussions

Within the planning sphere, the starting condition typically shows 3rd party domination accompanied by mere adoption and offer-oriented selection of goals while at the same time maintaining a degree of inquisitiveness (‘stubbornness’, ‘fun in learning’, ‘interest’, and ‘inquisitiveness’). The ability to explicate intentions as well as objectives and the ability to set priorities to tasks were found among these necessary triggers. Those are subject to be developed in order to support the acquisition of the ability to design one’s own portfolio and the ability to match formal with personal requirements where necessary.

With respect to reflection, willingness to change one’s attitudes was identified as the minimal condition to start building a PLE, whereas at the opposite end of the competence continuum, the ability to actively engage in the process of reviewing own (digital) traces, identifying strengths and weaknesses of one’s own performance and thus making sense of the past learning experiences seem to be of importance. In this case the trigger allowing the passage from one stage to another is the process of ‘leaving (digital) traces’ – producing, publishing, and collecting learning artefacts.

Looking at monitoring, the minimal condition typically includes such 'silver-spoon' learning habits as adopting external evaluation criteria and relying on the evaluation of performance on the more knowledgeable other (typically a teacher). Among the necessary triggers the ability to build up criteria for self-evaluation and willingness to ‘put yourself and your artefacts out [to the wider public] and get feedback by others, or read others contributions for measuring and comparison’ were identified. These attitudes and abilities are necessary to trigger the acquisition of the ability to control and direct learning progress in an autonomous and disciplined way, and to actively design and direct the relationship with the facilitator(s).

The next area, – acting –, encompasses a set of skills, attitudes, abilities, habits, and knowledge that are closely related to the concepts of Media and Digital Literacies. As a minimal condition for constructing a PLE a learner needs basic skills in ICT, and the ability to search for, collect and store relevant information. Another very important minimal condition in this sphere is the willingness to try something new. Among the relevant triggers the following competences (abilities and attitudes) from the interviews were identified: comprehension that information beyond printed media (such as Wiki pages, blog entries, or peer comments) has a high value for learning and knowledge

of 50

construction; ability to deconstruct one’s PLE i.e. the ‘capability to think about what comprises one’s personal learning environment’; ability to screen available tools and as a result make an informed selection; and, last but not least, the ability to re-purpose a tool, i.e. perceive affordances emerging from a new context.

The above-mentioned abilities and attitudes are important prerequisites for the acquisition of high levels of digital literacy. These intended high-level outcomes encompass: the ability to assess the quality and reliability of information and the ability to make productive use of technology, including matching the right tool for the right job. Furthermore, the construction and maintenance of a tool portfolio is now a competence fitted with a degree of creativity and framed by constantly enriched experiences not only with the tools therein but also with combined application of two or more tools. Whereas the minimal condition requires information gathering skills that are by nature content-centred, the intended outcome is an ability to regulate with respect to both content as well as the learning process.

Within the interacting sphere, basic social skills, a social interest, and particularly foreign language skills define the minimal condition. The triggers are mainly characterised by attitude changes: a willingness for exposure needs to be developed – a willingness such as ‘sharing artefacts with others’, ‘publishing not only consuming’, or – more general – ‘giving not only taking’. Additionally, extended social skills are required to set off the acquisition of the targeted competences. These extended social skills specifically are related to dealing with mediated communication and mediated criticism, often underestimated in their impact on performance. Intended outcomes were identified to be networking competence including the ability to network for feedback, decision competence on when to work in a group, and negotiation skills. The latter is related to reaching agreement within a group with respect to, e.g., roles, rules, and tools.

2.3 Summary on strategic competences in connection with PLEs The case study led to a competence spectrum along five initial dimensions. The PLE competence continuum consisting of skills, abilities, habits, and attitudes can be partitioned into three stages guiding from the situation at the beginning to the intended outcome. These preliminary results indicate that certain competences are required for and developed through PLE-based activities. Although the PLE competence continuum clearly shows that basic instrumental and informational digital literacy skills are a prerequisite – ‘minimal condition’ – for starting to consciously build a personal learning environment, PLEs should effectively support the development of higher level (strategic) digital and media literacy skills.

A probable reason for this is that monolithic applications (such as virtual learning environments, also known as learning management systems) used in education are outdated and do not promote the development of relevant digital literacy skills. It seems that the PLE concept incorporates more recent IT approaches and a more challenging truly distributed setting that both require the learner to take a more active role in managing and configuring the involved systems. PLEs can be said as a means to promote digital literacy in a spiral development process if a user has achieved some

of 50

basic skills and competences on PLE usage before. Hereby, good practice sharing can be an enabler for supporting this spiral development of media and digital literacy skills.

The development of PLE-related competences is triggered by interactions with the learning environment. Recordings of learner interactions and, in further consequence, activity patterns scaffold people into developing both professional competences and transcompetences. Besides, patterns implicitly include actors with their roles, their competences, and even user profiles. Thus, patterns might be described by users, e.g. through a minimal set of metadata like the learning context and the role of the actor, while it would not be helpful to model and assess every possible competence identifiable within an activity.

Overall, interaction recordings and patterns allow analysing characteristics of networked communities, certain learner behaviour and the underlying competences, which will also be addressed in the two upcoming sections. Analysing these theoretical considerations on professional and transcompetences from a more practical level, the upcoming section elaborates how activity patterns which include these competences implicitly can be captured and shared within a learning community.

of 50

3 Capturing and sharing activity patterns Next to PLE-related competences, the learning environment itself is the key for good practice sharing strategies. The materialisation of learner interactions with the environment and particularly the user’s PLE design decisions leads to digital recordings of learning experiences, as shown in the ROLE Deliverable D7.1/ID7.2 (“Model and Methodology for PLE-Based Collaboration in Learning Ecologies”). Before these interaction recordings can be shared, activity patterns have to be derived e.g. by removing or masquerading specific parts.

In the following, strategies and facilities from well-known research projects and software products are reviewed in order to find characteristics and key features relevant for capturing and sharing activity pattern. Particularly, this technology survey elaborates what can be captured and shared as a pattern and how the distillation and abstraction of good practice patterns are achievable. Consequently, this section highlights the characteristics relevant for activity pattern sharing and summarises strategies and facilities given by the review of state-of-the-art technology.

3.1 State-of-the-art overview An analysis of the interactions of (lifelong) learners (cf. section 2 and 3 of the ROLE Deliverable D7.1/ID7.2) shows that a learning ecology can be seen as a network in which actors are connected to agents, artefacts, communities, media, and processes (see Klamma & Petrushyna, 2008). From the perspective of a learner, the environment comprises everything of that learning ecology but the learner. In particular, the learning environment includes all entities a learner can directly and indirectly interact with:

• Processes: Lifelong learning activities carried out at the workplace, for educational reasons, or due to personal goals (e.g. a job task in a business process, attending a course for further education, or a spare time activity requiring the acquisition of new competences)

• Media: Collection of learning resources required for or created in these activities (e.g. the Wikipedia platform, learning objects repository, or simply the Internet)

• Artefacts: Documents and other (digital or real-world) entities learners collaboratively work on (e.g. a Wiki article or a joint paper)

• Communities: People sharing the same environment, e.g. in terms of having common interests, working on the same artefacts, being connected to the same actors (e.g. a group of learners trying to achieve a course goal or a special interest group for a specific topic)

• Agents: Other actors, no matter if human or systemic ones (e.g. peer learners or functionality provided via Internet)

of 50

The learning environment, therefore, includes all the relations between these entities, whereby the interactions of actors with artefacts and other actors are of particular interest for PLE-based collaboration (see also ROLE Deliverable D7.1/ID7.2).

Pattern Sharing

Sharing such patterns of user interactions, as defined with the good practice lifecycle in Figure 1-1, can be identified in various software solutions and in different forms. Appendix A.1 summarises a review of technological approaches. The survey is conducted along six categories, namely (1) tracking and recording services, (2) personal and mash-up websites, (3) widget engines and learning management systems, (4) collaboration platforms, (5) workplace learning technologies, and (6) personal learning environment (PLE) solutions.

Each of these categories of technological systems includes aspects or the full palette of the pattern sharing lifecycle. For each category approaches are selected according to their relevance, popularity, and user-friendliness. Besides giving an overview of each solution, it is explained how good practice sharing is realised. The detailed description and analysis of the pattern sharing approaches can be found in Appendix A.1. At the same time, this technology survey manifests key features for a pattern-based practice sharing solution which are summarised in Appendix A.3.

As a short conclusion of the technologies reviewed in Appendix A.1, it can be manifested that learner tracking can be realised based on different aspects, like the possibilities to structure the learning context, the kinds of interactions and applications to track, the social form, or the architecture of the pattern capturing infrastructure. Due to trust and privacy concerns, the user should have full control over the sensitive interaction recordings. Before sharing, specific parts of the recordings should be de-personalizable, so that learners can remove or masquerade personal and sensitive data. Then, other users who want to use these patterns should be able to re-personalize the recordings again in order to use the shared learning experience to the full extent. In addition, the creation and sharing need to be doable by users with low digital literacy.

Finally and mostly relevant for a community approach, it requires a repository where patterns can be stored by learners and thus are searchable and reusable others. Regarding pattern repositories, section 4 elaborates this special aspect of activity pattern sharing. In the following, the different ways how patterns are stored in existing approaches are examined.

Pattern Capturing

After having identified the approaches for pattern sharing from rather different fields, this subsection also reviews formalisms which can be used to materialise activity patterns. Such data formats are normally not standardised or, in the case of commercial service providers like Mendeley, not even public. As a consequence, the technology survey summarised in Appendix A.2 presents only a few selected approaches which aim at capturing activity patterns. Again, the state-of-the-art analysis of pattern capturing formalisms is conducted along the six technological categories of systems which realise aspects of pattern sharing.

Summarising this technology survey, there is no formalism which meets all requirements for patterns of PLE-based, collaborative activities. Either pattern sharing is based on generic macro languages missing aspects of activity structures and learning situations, or approaches do not support de-personalization or group activities. Additionally, the storage location of the interaction recordings seems to be an important issue; an offset has to be made between client-sided solution (i.e. to ensure trustworthiness, preserve one’s privacy, and consider local application) and a server-sided platform (i.e. to capture group activities and share patterns). Consequently, analysis and synchronisation techniques are necessary for providing social recommendations, supporting collaborative activities, or analysing community aspects.

Again, the key features identified in this survey are listed in Appendix A.3. The surveys also indicate that there are specific characteristics – technical and non-technical ones – which are relevant for pattern capturing and sharing. Therefore, the following subsection summarises these dimensions and explains them in detail; then, subsection 3.3 highlights possible strategies and the key features gathered from the technology review.

3.2 Dimensions of pattern capturing and sharing Referring to the survey in the previous subsection, Figure 3-1 gives an overview of aspects which are relevant for capturing and sharing activity patterns: (1) architectural design, (2) activity structure, (3) interaction types, (4) tracked applications, (5) social form, and (6) de-personalization.

Fig. 3-1: Dimensions of capturing and sharing activity patterns in PLE settings

Each dimension of pattern sharing will be described and discussed in the following.

of 50

of 50

Architectural design

The first important dimension deals with the architecture of the pattern capturing and sharing infrastructure. As shown by existing approaches such an infrastructure can be located on the client-side (iMacros or Selenium), on the server-side (iGoogle, Netvibes, Google Wave, Elgg and the widget-enabled Moodle, or MUPPLE), or distributed between server and client. In the latter case, the client contains a small tracking application which sends the interaction recordings to an online service (Mendeley, Audioscrobbler, Wakoopa or APOSDLE). A platform for capturing and sharing activity patterns can be also closed, restricting the functionality to the software package delivered (like the APOSDLE system or a default Moodle installation), or open, i.e. allowing end-users to add widgets and content from external sources (such as a widget-enabled Moodle platform, Elgg or typical PLE solutions).

The tracking data can be either submitted fully to the online API (Wakoopa, Mendeley) or in a compressed form, e.g. by deriving inferences automatically (APOSDLE, MUPPLE) or by explicit approval by the user (Shareaholic, MUPPLE). Due to trust and privacy concerns, it is beneficial to give the user control over the sensitive data and provide facilities to actively share activity patterns over a central web service, e.g. through an own API of a practice sharing server (iMacros, CAM). On the other hand, specific functionality, like capturing activity patterns of groups, requires access to the interaction recordings in different systems (PALADIN).

Activity structure

Another dimension comprises the underlying pedagogical model, which has to be generic in order to be supportive for the user’s online activities. As elaborated in the ROLE Deliverable D7.1/ID7.2, this model should be generic enough so that the very broad range of lifelong learning activities can be described. With respect to experiences from the MUPPLE prototype, it should not be based on complex instructional design theories, as it is intended also for non-educational and non-computer experts (Mödritscher & Wild, 2009). The model should also allow flat and hierarchical activity structures, maybe also taking into consideration simple interaction flows but without including the complexity of instructional design models.

Overall, the pedagogical model should support many kinds of lifelong learning activities. Similar to other approaches like MUPPLE or the INCENSE system, the ROLE Deliverable D7.1/ID7.2 proposes such a model based on Activity Theory and demonstrates how everyday activities from the ROLE test-beds can be structured with it. Moreover, instructional design languages might be appropriate for the purpose of structuring the contexts of lifelong learners if the complexity of instructional modelling (cf. Derntl et al., 2009) and even its terminology are hidden away from the end-users.

Interaction type

Highly relevant for PLE-based collaboration is the users’ interaction with (web-based) tools and shared artefacts so that learners can achieve their personal goals. The tracking and recording services reviewed in the last two subsections comprise tools for capturing and replaying these learner interactions. On the one hand, iMacros, Selenium,

of 50

Wakoopa, Audioscrobbler, etc. allow recording interactions with the browser using a pre-defined language which enables users to specify the URLs. On the other hand, Shareaholic, MyYahoo (and Pipes), Google Wave, etc. support user-given input in the form of metadata and content.

Therefore, the dimension of interactions types can be spanned from pre-defined interactions to interactions (partially or fully) entered by the users. Various widget engines and PLE solutions provide evidence for the trend that end-users should not only generate and describe content artefacts but also select and tag the tools for their activities (Elgg, Moodle, Wookie). For instance, APOSDLE supports the integration of the organisational repositories which include all the artefacts created or managed by the knowledge workers into their own platform (cf. Mödritscher et al., 2007), while MUPPLE even forces learners to name the action and the goal (outcome) of each tool used. The MUPPLE approach, however, aims at changing the learning behaviour slightly in order to achieve that learners (and not the experts) determine the purpose of each interaction.

Tracked applications

Another dimension of capturing and sharing activity patterns addresses the type of applications considered by the materialisation of learner interactions. Tracking services and widget engines, as examined in the state-of-the-art part of this section, mainly focus on interactions with web environments, whereby only Wakoopa and RescueTime include desktop applications. In fact, also PLE platforms are based on server-sided architectures, thus ignoring the tools installed locally on the computers of the learners.

An approach for monitoring user behaviour from the server-side is the MobSOS success model described by Renzel et al. (2008). The purpose of MobSOS is to define best practices based on user activities followed by questionnaires. The activities of users that operate with different services hosted by a server are tracked from the server-side and then can be categorised and arranged in patterns of learners. Concerning activity patterns, the dimension of tracked applications reaches from local applications to web-based tools.

Social form

A fifth dimension of activity patterns deals with the question how experiences of single learners or whole group activities are materialised. With respect to client-sided tracking and storing of interaction recordings (iMacros, Selenium), it is not possible to reconstruct group activities. Capturing user interactions on a server (MUPPLE, widget engines, Google Waves) or providing the possibility to share recordings over a web-based service (Shareaholic, iMacros) allows analysing all the interaction recordings in order to create patterns of group activities or, like in the case of the PALADIN approach, to detect network effects and specific user roles within a community. The dimension of the social form comprises patterns of single actors, groups, and the community.

De-personalization

Digital recordings of the interactions of learners with their environment contain sensitive data, like URLs to personal or organisational material or interactions that a user does not want to share with others. In his user-centric privacy framework with the special

of 50

focus on open systems, García-Barrios (2009) introduces a strategy to raise the users’ awareness about privacy-critical features and content, i.e. through visualising the benefits achievable within the environment in relation with opening up one’s data. However, this framework includes also user-adaptable privacy statements to protect personal data in an open ICT infrastructure.

Drawing conclusions from this user-centric privacy framework, digitalised recordings of learner interactions should not be shared without bringing privacy and trust issues to the learners’ attention and without providing facilities for securing one’s data. Precisely, these recordings could include personal or restricted information, e.g. URLs to a sensitive document of a company or a collection of multimedia content being protected by copyrights. Therefore, it must be possible to de-personalize learner interaction recordings before sharing them in social networking platforms.

Different from the psychological malfunction ‘depersonalization’, this term is used for the process of removing or masquerading sensitive items in digital recordings of learner interactions. Particularly driven by social networking approaches, anonymization is necessary to preserve the user data in social sites (Zhou et al., 2008). Besides anonymizing the user-generated data, de-personalization also considers sensitive data collaboratively created by groups or within a company. De-personalization can be achieved by fostering user awareness, manual facilities (e.g. by the placeholder mechanism of the MUPPLE approach) or fully automated approaches (e.g. by mining techniques in APOSDLE and Wakoopa).

These six dimensions can be used to evaluate existing approaches and formalisms for pattern capturing and sharing. In combination with the state-of-the-art analysis of this section, they can be also used to discuss strategies and facilities for activity pattern sharing, which is done in the following subsection.

3.3 Summary on state-of-the-art strategies and facilities In this section, two important questions of pattern sharing were addressed. On the one hand, state-of-the-art approaches in which pattern capturing and sharing can be identified were reviewed to examine what kinds of interactions are captured and shared. On the other hand, this section also manifests concrete formats used to materialise user interactions with PLE-like environments. Both technology surveys lead to the conclusion that specific aspects are missing.

Trust and privacy concerns as well as state-of-the-art technology indicate that capturing user interactions should be part of the PLE solution and ideally be realised on the client-side while practice sharing requires server-sided services. Learners should be supported in structuring their lifelong learning activities through flat and hierarchical activity models, probably also including elements for defining simple action flows. Furthermore, a practice sharing infrastructure should consider pre-defined and user-specified interactions with web and desktop applications used by learners.

Moreover, patterns of PLE-based activities can be generated for single users and groups, while such recordings also allow analysing characteristics of networked communities and, thus, provide additional facilities for learning communities. Finally and

of 50

referring to privacy considerations again, de-personalization of interaction recordings was identified as a critical requirement to secure sensitive user data.

Possible strategies and facilities

All in all, the work on such a practice sharing infrastructure has already been started. Concerning client-sided tools for capturing interactions with the learning environments, three prototypes are being developed by January 2010.

First of all, the Vienna University of Economics and Business is working on a personal activity management tool, as sketched in the ROLE Deliverable D7.1/ID7.2. This Firefox add-on enables learners to design their activities and manage the tools they need within these activities. Secondly, the Open University UK is participating in the JetPack competition and building a prototype named ‘MUPPLE II’ which aims at capturing and replaying web activities. Both developments are described in the internal ROLE deliverable ID7.1 (“Draft Prototype of a Mash-up PLE”). Thirdly, outside the ROLE project, the Bamboo Learning prototype being developed by the aDeNu Research group at the Spanish National University for Distance Education (UNED) comprises another JetPack solution for learning network management (see http://adenu.ia.uned.es/web/en/projects/jetpack_learning_design_challenge). Again, this client-sided prototype provides facilities to record learner interactions in the background and to browse and replay these recordings.

Server-sided PLE solutions, on the other hand, are being developed with different technologies and in different areas. The state-of-the-art evaluation in subsection 3.1 included already various platforms (e.g. Wookie, widget-enabled Moodle, MUPPLE, Elgg, etc.) which can be understood as a server-sided environment supporting interaction tracking. Similar to social networking sites like Facebook, pattern sharing strategies can be realised through automated mechanisms (i.e. recommendations) or facilities for manually sending patterns.

Work on formalisms to capture and share PLE-based learning experiences is in its early stage. Amongst others, the ROLE Deliverable D7.1/ID7.2 reported about a scripting approach for this purpose. Next to this scripting language, Google Wave materialises the content artefacts which are collaboratively created by several users through a preliminary json object, while widget engines and personalized websites store mash-ups with XML-based description formats. Furthermore, the MUPPLE II prototype materialises learner interactions in the form of RDFa resources on the local computer. The PALADIN approach stores the network structure retrieved from the log files of various systems into a database.

Beside the key features which have been identified in the technology evaluation and are summarised in Appendix A.3, the most important conclusion of this section is that a repository is required for both aspects of pattern sharing, for manual approaches as well as for automated ones like recommendation mining. Therefore, the next section elaborates possible strategies and facilities in connection with pattern repositories.

of 50

4 Pattern Repositories In accordance with experiences from approaches like Audioscrobbler, Wakoopa or Mendeley, pattern sharing in PLE settings requires specific services through which learners can offer de-personalized recordings of user interactions (the activity patterns), and peers can find these user experiences and reuse them for their own purposes. The patterns are shareable if users can remove or masquerade sensitive parts of their interaction recordings. In return, users who want to make use of a pattern might personalize the missing information, e.g. by entering it again. By instantiating a pattern, a learner should get a PLE pre-configured for a specific activity. Moreover, a good practice sharing service enables automated techniques like providing useful recommendation if an adequate number of recordings have been submitted.

In the scope of this document, such an online service is called pattern repository. Such repositories have an open API for publishing and retrieving digitalised interaction recordings. In accordance with the lifecycle of good practices (see Figure 1-1), the sharing process consists of the de-personalization activity by which a user removes or masquerades sensitive data and the publishing of the patterns on a shared space. Furthermore, the retrieval process includes aspects of providing the activity patterns, e.g. by browsing or search functionality, or of recommendation mining. Finally, instantiation of a pattern might require manual personalization steps.

The next section elaborates state-of-the-art concepts and technology for pattern repositories. Thereafter, existing approaches and platforms are reviewed according to their applicability as a pattern repository. Finally, the conclusions on pattern repositories are summarised in the form of strategies and facilities for the ROLE project.

4.1 State-of-the-art overview In the scope of good practice sharing, pattern repositories have to support two important processes: publishing and retrieval of activity patterns.

De-personalization and publishing of patterns

The last subsection evidenced that de-personalization of activity patterns is hardly given for manual sharing strategies, while the concept of anonymization has been well elaborated for social networking platforms (cf. Zhou et al., 2008), mainly to preserve the users’ privacy. Anonymization comprises methods to achieve anonymity, i.e. by removing person-related information by which the identity of a user can be tracked back. Anonymization normally is applied to sensitive areas, e.g. health or bioinformatics to remove patients’ data from tissue samples (Eder et al., 2007). In the scope of social networking, anonymization can be achieved through authentication (anonymity or pseudonymity, cf. Pfitzmann & Hansen, 2008), as a part of recommendation mining techniques (k-anonymization or l-diversity, cf. Zhou et al., 2008) or through social network analysis approaches (Das et al., 2009).

of 50

However, user-controlled de-personalization can be hardly identified in practice. One related approach is Google Docs (http://docs.google.com) which allows sharing of online documents, i.e. requiring users to edit the document with an in-line editor before it can be shared as template. Moreover, Yahoo Pipes (http://pipes.yahoo.com) includes a form-based mechanism to masquerade specific user input before a user-generated pipe can be shared to others. Beside anonymization techniques, such methods are necessary to distil sharable activity patterns from the full interaction recordings.

In the scope of online communities and social websites, publishing comprises the process of disseminating information by users. Particularly, the Web 2.0 stream aims at creating online spaces for active content contributors, the so-called read-write Web (O’Reilly, 2005). Publishing platforms are typically blogs either integrated in one’s personal websites or provided by social networking sites like Facebook and MySpace. These kinds of publishing approaches are intended for content artefacts (textual information, links, videos, etc.) from users for other users.

As mentioned in the last section, practice sharing requires ‘recordings of good practices’ which are subject to de-personalization and publishing. Thus, such a pattern repository has to provide a specialised API for publishing interaction recordings from the clients, similarly to the Last.fm API which is fed by the Audioscrobbler clients (see http://www.lastfm.de/api) or the Simple Publishing Interface specification (http://www.cen-ltso.net/Main.aspx?put=1048). Optionally, such a pattern repository could have a user interface with statistical information and recommendations on the stored patterns.

Retrieval and instantiation of patterns

The more important aspect of a pattern repository is the one about finding and instantiating activity patterns.

Referring to information retrieval strategies, can include facilities for actively browsing and searching the patterns of a repository. However, Resnick & Varian (1997) state that recommendations are necessary if users have to make choices without sufficient personal experiences of alternatives, which is considerably the case for lifelong learners who try to utilise PLE technology for their very different learning contexts. In the sense of recommender systems, activity patterns derived from interaction recordings can be analysed according to specific aspects, which might be worth recommending to other learners. However, mining and providing recommendation should neither threaten the users’ privacy nor decrease the trustworthiness of the PLE infrastructure, as evidenced with a study on trust and privacy concerns in social networks (Dwyer et al., 2007).

Amongst others, current community approaches like Last.fm or Wakoopa offer recommendations on music sequences to listen to in a specific situation, desktop applications to use for a certain task, or scientific papers to read for the topic a researcher is currently working on. Thereby, these community-based solutions offer a fully open API (e.g. the Last.fm API), an open API requiring authentication (e.g. the Wakoopa API) or a hidden one (e.g. commercial products like the first version of the Mendeley API which is about to be opened up).

of 50

Klamma et al. (2009) report on a recommender system for scientific communities AERCS based on paper writing and co-author citation activities. The model of AERCS uses Social Network Analysis for scientific community to understand the structure and pattern of research collaboration. AERCS builds an author network within a conference or journal, computes similarities between target researchers and generate recommendations based on activities of users with a similar behaviour (e.g. the participation in conferences or their publications in journals). A recommender system like AERCS is particularly useful for young researchers to find academic partners to collaborate with as well as conferences and journals to submit papers.

In the scope of collaborative learning platforms, El Helou et al. (2009) proposes an approach which is based on a model of actors, assets, and group activities (3A model) and ranks these three entities according to the target user and her current context. In a concrete example, the contextual and multi-relational ranking algorithm introduced by El Helou and her colleagues has been applied to analyse the deliverable writing process in an EU project and is intended to support such activities, i.e. by recommending actors (with different roles), appropriate assets, or activities from related situations.

A pattern repository could support users while working on a lifelong learning activity in two ways. On a macro level, it can provide pre-configured PLEs in the form of activity patterns before the activity is started. On a micro level and while being involved in an activity, it can support learners in adapting their PLEs by recommending specific tools, certain documents, or relevant peer learners. Therefore, the 3A contextual ranking system introduced by El Helou et al (2009) re-calculates the rankings based on the changes in the database. These ranked lists of actors, assets, and activities could be provided to end-users through information pull and push strategies. On server-sided platforms such approaches can be identified at social networking platforms like Facebook (http://facebook.com) or the MUPPLE prototype (Wild et al., 2009). Concerning other community platforms, Last.fm offers song recommendations over its API during listening to the music (micro level), while Mendeley plans to inform scientists about new developments actively (macro level).

Besides retrieving activity patterns (through information push or pull mechanisms), such patterns also have to be instantiated before they can be used. Instantiation, therefore, is the process of initialising and (re-)personalizing the environment which has been recommended in the form of an activity pattern. Dependent on the de-personalization technique applied, this instantiation can be realised through facilities for specifying removed or masqueraded parts of the activity patterns. For instance, the MUPPLE prototype requires end-users to enter the parts of a pattern which have been de-personalized with placeholders by the learner sharing this pattern (Wild, 2009). Yahoo Pipes (http://pipes.yahoo.com), on the other hand, prompts users to initialise the fields required for launching a tool mash-up. PALADIN allows users to define the patterns based on different social network properties that are prerequisites of actions (communicative person, silent person).

Nevertheless, an approach based on pattern sharing requires end-users to initialise the learning environment by re-personalizing the recommended pattern, before they can

of 50

use these PLEs. In the following subsection, existing platforms are discussed in terms of their appropriateness for being a pattern repository, as elaborated before.

4.2 Related work and discussion Regarding existing pattern repository technology, the following paragraphs describe examples of pattern sharing technology throughout various communities and discuss their appropriateness for the purpose of good practice sharing in PLE settings.

A first example of a pattern repository is the music community site Last.fm (http://last.fm) which provides the world’s largest online catalogue with a lot of information on music. Next to the catalogue service, Last.fm also provides recommendations which are based on users’ listening behaviour captured through the Audioscrobbler application (see Appendix A.1 and A.2). The pattern repository is realised in the form of an open API which allows uploading user interaction recordings and retrieving information on albums, artists, events, groups, libraries, radio playlists, tracks, and user feedback. Hereby, Audioscrobbler uploads the user interactions, yet without any kind of de-personalization of the recordings. The platform itself analyses these recordings and ranks them according to the recommendation categories Last.fm is providing. In summary, this repository seems to be a first good approach for a good practice sharing repository. However, the main shortcoming for a PLE solution is that it lacks de-personalization techniques, thus not being a real ‘pattern’ repository and not preserving learners’ privacy sufficiently.

Secondly, e-teaching.org (see http://e-teaching.org) is another community site which provides best practices for selected aspects of teaching. The practices are created and published by didactical experts, trying to address the target group of teachers and online learning vacillators. In contrary to Last.fm, however, e-teaching.org cannot be seen as a real pattern repository. The practices are anonymized and open to all users, and they include relevant links and descriptions of web-based tools in the Internet. Nevertheless, publishing is restricted to a very few users, the experts. Moreover, this practice sharing platform does not consider user behaviour or user interaction at all neither does it provide recommendations which are appropriate for specific needs and contexts of users.

Thirdly and more related to end-users, the so-called Pattern Language Network for Web 2.0 Learning (PLaNet, cf. http://patternlanguagenetwork.org) comprises a project which aims at sharing educational strategies with Web 2.0 technologies. In this online community, all kind of users can share experiences with content creation and tool usage in everyday education. The scenarios are described through specific metadata fields and materialised with PLML (Pattern Language Markup Language). In principle, the PLaNet idea comprises a pattern repository which includes aspects of collaboration, the utilisation for tools and practice sharing through patterns. Moreover, it is also possible to de-personalize interaction recordings from sensitive data. Similar pattern collections are available by the Patterns Library of Hillside.net (http://hillside.net/patterns) or Pattern Language Network (http://patternlanguagenetwork.wordpress.com). Yet, these approaches lack a community platform as well as a focus on experiences of learners.

of 50

In the scope of the CEWebS project (Cooperative Environment Web Services), Motsching-Pitrik & Derntl (2005) describe conceptual modelling of technology-enhanced learning patterns in organisations and through a pattern repository. Web-based activities of learners are arranged in patterns that are hosted in a pattern package like assessment, course type, evaluation, feedback and so on. The patterns are defined using UML static structure so that inter-relations of patterns can be specified. The described patterns of activity sharing are only course-specific and do not consider other models of learning that take place within PLE environments.

The Kaleidoscope project (see http://www.kaleidoscopeproject.org.uk) focussed on the usage of a learning patterns repository for the TEL community. The patterns indicate how to use resources in everyday practice. Particularly, Kaleidoscope patterns deal with the design, the development and the deployment of games for mathematical learning. It includes an organised structure of pattern design process and deployment as well as reflective mechanism of sharing individual and social experiences of pattern usage. The patterns advise different game elements for different pedagogical purposes and different kind of students as well as explain particular behaviours of students (missing the points, getting bored, etc). The patterns, however, are created and provided by experts, i.e. as an outcome of the project.

Similarly, the Technology-Supported Learning Database (TLSDB) of the Edith Cowan University (http://aragorn.scca.ecu.edu.au/tsldb) offers educational activities which are comparable to activity patterns. These activities can be described with metadata, browsed and searched. An activity includes all information, i.e. the roles, the expected outcomes, hardware and software requirements, etc., necessary for implementing it in classroom or online teaching and can be rated and commented by the community. Nevertheless, the TLSDB does not contain user interactions or good practices from learners. This pattern repository is more a platform for teachers who want to share and publish successful teaching activities.

An approach dealing with capturing and sharing learner interaction recordings is LAMS which stands for Learning Activity Management System. LAMS is a tool for designing, managing and delivering online collaborative activities, originally intended to be for teachers but then extended by a template (pattern) mechanism and facilities for learner-driven activity design (see Figure 4-1). So far, LAMS provides an authoring tool for comfortably designing so-called ‘templates’ which are conceptually similar to the activity patterns defined in this document. The LAMS approach, anyhow, consists of a web-based platform and client-sided tools, whereby LAMS activities and templates can be imported into learning management systems. The full functionality of the templates is provided by the LAMS platform only. Furthermore, this template approach rather aims at providing and monitoring learning experiences and does not include aspects of automated pattern sharing e.g. through recommendations at all.

Fig. 4-1: Example LAMS template for scientific streams, taken from Dalziel (2007)

The LAMS community even provides a repository of activity sequences (patterns) through an own platform (http://lamscommunity.org/lamscentral). Again, the patterns are described with metadata and can be retrieved trough browsing and searching facilities. In contrary to other pattern repository like PLaNet or TLSDB, this repository partially contains real interaction recordings in the form of LAMS packages of the activity sequences. So far, LAMS is closest to the concept of pattern sharing, although it is not intended to be a backend service for PLE solutions. The main disadvantage of the packages is its dependency on the LAMS software thus being not open to other learning solutions if the full functionality of the patterns is required.

Another pattern repository can be identified in the research project APOSDLE which aims at developing a software platform and tools to support workplace learning (see http://www.aposdle.tugraz.at). In the third prototype (see APOSDLE, 2009), the APOSDLE solution comprises a set of learning and working tools as well as a platform with a web service API for communicating with the client applications. Amongst others, the client applications provide information about a learner’s working context, her competences or the tools used. The platform then stores and analyses this data and provides recommendations on artefacts, learning events, and experts helpful for a specific working task.

Manual practice sharing in the form of patterns can be achieved by context-sensitive learning events whereby de-personalization is realised by reducing sensitive user data to so-called learning goals, i.e. a domain-independent taxonomy of educational activities. The learning events from the goal-based patterns by instantiating the topics relevant for a working task automatically. Consequently, the patterns are enriched by domain-specific content artefacts which relate to the learning context of the end-user. Overall, the APOSDLE platform is considered to be what is called a pattern repository in this document. Due to the focus on workplace learning, the main shortcoming of APOSDLE deals with the fact that the approach builds upon a closed architecture,

of 50

of 50

including a pre-defined set of tools. Furthermore, the materialisation of practices is reduced to templates for learning events and competency models stored within a database, which hinders manual sharing amongst users. Thus, this platform is not supposed to be applicable for a community approach.

4.3 Summary on repository strategies and facilities In summary, pattern repositories are available already in industrial and research practice, although they have different labels or are part of larger approaches. A key feature seems to be an open, specialised API by which different kinds of (client- and server-based) PLE solutions can publish and retrieve recordings of practices. Publishing should ensure trustworthiness and create awareness of privacy issues, e.g. warnings that personal information is accessible through others or even the visualisation of the sensitive data concerned (see García-Barrios, 2009), and provide facilities for enabling learners to de-personalize interaction recordings. Additionally, a pattern repository has to support metadata descriptions of the digital pattern artefacts, such as CAM, for describing the usage context of tools.

Retrieving good practices from such a repository can be realised through different channels. Besides classical facilities for browsing and searching, the generation of recommendations is an important issue. Recommendations can be offered through ranked lists of specific entities in PLE-based activities (peers, artefacts, activities) or information push strategies for such aspects. Thereby, the repository could offer a pre-configured PLE, i.e. the full activity pattern, before starting a specific situation or just single interactions in the form of a tool while learners are working on one of their activities. Re-using learning experiences, learners should be also allowed to personalize the retrieved patterns before starting or while actively working on the activity.

As shown with the technology evaluation in this section, the pattern repository solutions available do not fully support the key features identified, as pattern platforms either contain only expert-given content (e-teaching.org, PLaNet, etc.), are based on design models requiring specific expertise (LAMS or any IMS-LD based approach), have no focus on learning at all (Last.fm, Wakoopa, RescueTime), or are closed systems not being applicable for a community approach (APOSDLE).

of 50

5 Summary on possible strategies and facilities Concluding this deliverable, the findings of the case study and technology surveys conducted are summarised, and a strategy for a PLE-based community approach, as targeted by the ROLE project, is proposed. Furthermore, the usefulness of the proposed strategy towards existing PLE solutions is discussed, and next steps are indicated.

5.1 Conclusions of the case study and the technology surveys Regarding competences required for PLE-based collaboration, section 2 indicated that learner interaction recordings as well as activity patterns include both professional and transcompetences. Without attempting to model and assess these PLE-related competences, they can be supported by (user-given) metadata for a better retrieval of good practices in the form of patterns and by providing specific facilities for fostering transcompetences like environment design capabilities or self-regulated learning skills. Patterns, furthermore, are useful for further analysis on the underlying competences of an activity, which consequently can lead to other support facilities.

Recapitalising section 3, pattern capturing and sharing can be characterised along architectural design issues, the possibilities to structure the learning context, the kinds of interaction types, the kinds of applications to track, the social dimension, and de-personalization strategies. Ideally, a PLE considers pre-defined and user-given learner interactions, local and web applications, single user and group activities, as well as community effects. Furthermore, de-personalization has been identified as an important concept to distil and abstract patterns from interaction recordings which can contain sensitive data, either by automated mechanisms or by facilities for manual editing.

Addressing the software architecture, a PLE solution which is supposed to capture learner interactions can be realised following different architectural designs, either as a fully server-sided application, like Graaasp, MIUPPLE, or Facebook, or as a PLE client application. Concerning trust, privacy, and responsiveness, the second variant seems to be the better one, as the sensitive data is located on the user’s computer, and local applications are considered. However, both architectures and all PLE solutions available have their place in practice, as it is open to the end-user which PLE she prefers.

Next to the PLE solution, good practice sharing requires a repository (see section 4) which enables users to publish and retrieve activity patterns, both realisable through a public API. The pattern repository can be part of the PLE solution, as seen for the MUPPLE prototype, or it can be implemented as separate online service. A pattern repository can also be dedicated to a specific context, e.g. good practices for work-integrated learning in a company, for a special interest group, for an educational institution, etc. Moreover, it can be accessed by server and client-sided PLE solutions.

In sum, a pattern repository should provide facilities for creating privacy awareness and for de-personalizing interaction recordings to be published. Furthermore, the public API should also offer ranked lists of relevant entities, like activity patterns for specific situations, relevant artefacts or tools for an activity, etc., as well as information push

features, such as notifications and recommendations. This information, then, can be requested by the PLE infrastructure which presents it to the end-user. As shown with repositories like TLSDB or LAMS, a web-based front-end with retrieval facilities as well as statistical information and pre-views of the learning experiences available would be a nice-to-have for a pattern repository.

5.2 Proposed strategy and facilities for sharing good PLE practices Based on this summary of technological aspects of good practice sharing in PLE settings, Figure 5-1 proposes adding one more component to the heterogeneous landscape of PLE infrastructures. This additional component is a pattern repository with a public API which provides methods for publishing and retrieving activity patterns by the PLE solutions.

Fig. 5-1: Proposed architecture of practice sharing infrastructure

The example architecture in Figure 5-1 shows that this approach allows practice sharing over different PLE implementations, e.g. client or server-sided solutions. Furthermore, pattern repositories support learners in sharing and retrieving user experiences on PLE-based learning situations. In addition to that, it is also possible to have more than one pattern repository and allow learners to integrate their preferred and required ones into their PLEs.

Capturing learner interactions should be part of the PLE solution itself, whereby client and server-sided implementations seem to have advantages as well as disadvantages. On the one hand, a client-sided PLE solution leaves the interaction recordings with the learner giving her full control over this sensitive data and thus increasing the trust into the PLE. Furthermore, the responsiveness of a client-sided PLE might be significantly higher as no Internet connection is required, and the usage of desktop applications can be tracked. On the other hand, server-side PLEs allow analysing user performance, networked collaboration, and community effects. Moreover, a PLE platform can already include pattern sharing strategies without using a separate pattern repository in the background, as evidenced with the MUPPLE prototype.

of 50

of 50

De-personalization might be realised through the PLE but can be supported by the pattern repository, e.g. in form of creating awareness about the data a learner is supposed to share. Due to interoperability reasons, publishing a pattern should be done in a uniform data format, requiring the transformation of other interaction recording formats into this one, as reported e.g. for the ARIADNE repository (cf. Najjar et al., 2003). As patterns can be added to a repository, rankings, recommendations, and social network analysis metrics need to be re-calculated regularly.

On the other hand, a pattern repository can optionally provide a web-based interface for browsing and retrieving patterns and support user comments, ratings, and additional statistical information on the data contained. Moreover, the public API has to include retrieval methods so that the PLE solution can access and offer the good practices stored at a pattern repository to the end-users. The retrieved patterns should be exportable in other formats in order to be instantiated by the PLE, so that they can be used by learners having different PLE solutions. The PLE has to handle the communication with the pattern repository. Therefore, a stable API considering downward compatibility is required.

The practice sharing architecture proposed in subsection 5.2 as well as the key features summarised in Appendix A are considered as input for the ROLE Deliverable D1.5 (“Revised requirements specification”). However, the evidence of the usefulness, the efficiency, and the effectiveness of such a pattern sharing infrastructure for PLEs is part of the future work, as it requires a PLE solution connected to at least one pattern repository which can be used for empirical studies with end-users in a real world setting like the ROLE test-beds (cf. Deliverable D5.1, “Design of the Personal Learning Test-beds”).

5.3 Applicability in PLE solutions and further action plan In the scope of the ROLE project, first PLE prototypes addressing this strategy have been identified already. The internal deliverable ID7.1 (“Draft Prototypes of a Mash-up PLE”) describes four PLE solutions being developed within the ROLE consortium. Two of these prototypes try to shift PLE functionality to the client-side, i.e. in the form of browser extensions (one XUL-based and one JetPack approach). Furthermore, the Spanish research group aDeNu is working on another JetPack tool for recording and replaying learner interactions (see http://adenu.ia.uned.es/web/en/projects/jetpack_learning_design_challenge).

Each of these client-sided solutions plus the two PLE platforms developed in the ROLE consortium (Graaasp and the Christmas Prototype) are suitable for being extended by an interface to a pattern repository, and being enriched by the possibility to share and reuse learning experiences of PLE-based, collaborative learning. As a next step towards a good practice sharing infrastructure for PLEs, the API for the pattern repository will be drafted and realised prototypically. Then, one of the PLE prototypes developed in ROLE should be plugged to this preliminary pattern repository.

of 50

Appendix A: Surveys of pattern sharing technology

A.1 Survey of technology for activity pattern sharing The survey of pattern sharing technology is conducted along six categories, namely (1) tracking and recording services, (2) personal and mash-up websites, (3) widget engines and learning management systems, (4) collaboration platforms, (5) workplace learning technologies, and (6) personal learning environment (PLE) solutions. Each of these categories includes aspects or the full palette of pattern sharing. For instance, tracking tools aim at materialising user interaction into digital recordings containing this user information, while personal websites and widget containers allow users to add functionality (widgets) to their sites. In the following, relevant, prominent and useful approaches for activity pattern sharing are examined along these six categories.

A first category of technical systems for capturing activity patterns can be determined with tracking and recording services. Typically, they can be characterised by recording user interactions on the client and storing the recordings either on the client or sending them to a server on the Internet. Amongst others, iMacros (a Firefox add-on, cf. http://www.iopus.com/imacros/firefox/?ref=fxmoz) enables users to record and replay interactions with their browser, thus allowing them to automate specific work flows. Similarly to a audio recorder, scripts of user interactions (so-called ‘macros’) with the browser can be recorded, edited, replayed, or shared with others via email or iMacros sharing service. Normally, the recordings are stored locally.

In order to support web application developers, the Firefox add-on Selenium (http://seleniumhq.org) provides a very similar functionality, except it has no sharing facilities. By recording and replaying user interaction sequences with the browser, developers can create test cases for their applications. Rather focussing on social networking, Shareaholic (http://www.shareaholic.com) is a plug-in for nearly all browsers, thus allowing users to manage their URLs over several bookmarking and social sites and structure their online activities.

Furthermore, certain applications track the user interactions on the client and send the data to online services in the background. Amongst others, Wakoopa (http://wakoopa.com) tracks which applications are used on a computer and, consequently, provides tagging and annotation functionality as well as tool recommendation according to tool usage patterns. In a similar way, RescueTime (http://www.rescuetime.com/) focuses on time tracking and management, thus helping individuals, project teams, and companies saving time in their everyday activities. Although not aiming at practice sharing, this approach which uses a web service to gather the user data provides recommendations for being more efficient in their activities as well as reflective elements like the “Quantified Self Blog”, a tool for analysing resource and tool usage.

Another approach is Audioscrobbler, a service provided by the music platform Last.fm (cf. http://www.audioscrobbler.net and http://www.last.fm) which tracks song listening habits of users and recommends new songs as well as song sequences based on statistical information about other users. More related to lifelong learning, Mendeley

of 50

(see http://www.mendeley.com and Henning & Reichelt, 2008) captures research activities on the user’s computer and provides facilities for organising one’s research library, sharing academic papers and practices with other scientists, and discovering new materials and trends.

A second category of technologies relevant for this survey comprises so-called personal and mash-up websites. In principle, these sites allow users to include and visually arrange web-based applications, i.e. widgets or portlets, in a portal-like framework, whereby certain providers also include aspects of sharing widget collections. Examples for personalisable websites are iGoogle (http:// www.google.com/ig) and Netvibes (http://www.netvibes.com) which allow users to integrate new widget from the space of compliant web applications available. These two platforms allow pattern sharing on the basis of widgets, so that users can include self-developed tools and actively recommend a widget to others.

Pageflakes (http://www.pageflakes.com) provides not only the technical infrastructure for social personalized homepages but also 235.000 ‘flakes’ (widgets) and 140.000 ‘pagecasts’ (pages) which are mainly developed and shared by users. A pagecast comprises the widget selection and arrangement for a specific purpose, designed by an end-user and shared with the community. Furthermore, flakes and pagecasts are recommended on the basis of tags and topicality.

Yahoo also offers a personalized website service, MyYahoo (http://my.yahoo.com), plus a mash-up service called Yahoo Pipes (http://pipes.yahoo.com). Using the online editor for pipes enables end-users combining widgets with each other and thus creating new functionality based on existing web applications. Furthermore, users can personalize these environments, e.g. by filling out form fields. This user-given data is submitted to and processed by each widget being part of the tool combination. The result returned by the widgets is visualised on a mash-up space on the screen. In such a setting, it is important that end-users have the possibility to remove or hide user-specific data before sharing learning experiences. In the scope of this document, the process of removing or masquerading personal data is called ‘de-personalization’.

The third category of technical solutions reviewed in this subsection includes widget engines and learning management systems allowing the integration and management of widgets. Examples of widget engines are Wookie (http://incubator.apache.org/wookie) or Elgg (http://elgg.org), whereby Wookie is more a development framework (Apache incubator stage) and Elgg aims at empowering users to create their social environment, including social networking and community features. In both cases, activity patterns are, like personalized websites, reduced to the selection and visual arrangement of widgets. Typically, widget engines also include functionality like a widget store by which users can browse and retrieve the widgets available. Within e-learning organisations the E-LEN project (http://www2.tisip.no/E-LEN) patterns are used as a means of communication, development, and dissemination of effective e-learning experiences. The other attempts of defining learning environment patterns were done in the pedagogical patterns project. However, the project does not explicitly address the use of learning technology (cf. Motsching-Pitrik & Derntl, 2005).

of 50

The open source learning management system Moodle (http://moodle.org) includes a plug-in which allows integrating and managing widgets, e.g. those who are provided by a Wookie server. Although not being part of educational activity flows, the widgets are at least visualised in the form of a course activity and thus being in the context of learning. However, adding and managing widgets are restricted to specific user roles. Furthermore, these activities are always based on pre-defined courses and therefore, neglecting aspects of situated learning or lifelong learning contexts. Although supporting learner-driven environment design, pattern sharing is very restricted in widget-based infrastructures and most personalized websites.

The next three categories of pattern sharing technologies are already highly specialised systems. Addressing collaboration platforms, Google is currently working on a platform for real-time communication and collaboration. Google Wave which is in the beta test stage (see http://wave.google.com) enables user groups to collaborate on a shared artefact (a so-called ‘Wave’), visualising the contributions of each user in an adequate way and recording the changes committed by the participants. Users can also integrate widgets into these group documents. Furthermore, users can replay the developmental history of a collaborative document and add or remove users at any time. Features for storing Waves locally or deriving new Waves from existing ones are planned, but pattern sharing has not been realised in this early development phase yet. Anyhow, this approach by Google perfectly realises the materialisation of learner interactions, as addressed by the ROLE Deliverable D7.1/ID7.2.

Regarding workplace learning technologies, the EU project APOSDLE (cf. http://www.aposdle.tugraz.at) has developed a prototype consisting of tools at the client-side and a server-sided platform. According to Lindstaedt et al. (2008), materialisation of learner interactions is achieved by a service on the computer which tracks all kinds of user input and compresses this information by attempting to determine the user context in terms of identifying tasks a user is involved in. Based on this information and a competency model located on the APOSDLE platform, the system provides tools relevant for the current work task and recommends learning events, documents, and experts for the knowledge learner’s situation. Hereby, pattern sharing is not realised by push strategies e.g. through facilities for manual pattern sharing but rather through automated mechanisms, like task determination, competence mining as well as content and expert recommendations.

Finally, personal learning environment (PLE) technology build upon the vision of empowering learners to design their own environments and thus require facilities for practice sharing (cf. Mödritscher & Wild, 2009). One solution approach developed within the EU project iCamp (http://icamp.eu) is the Mash-Up Personal Learning Environment (http://mupple.org) prototype which allows learners to create and manage so-called MUPPLE pages for each higher education activity (Wild, 2009). These pages are designed by the learners through drag-and-drop functionality leading to a visual mash-up of web applications for each activity. As different from personalizable websites, MUPPLE pages are based on a generic, action-oriented model of activities and require learners to specify the purpose of each web-based tool used in a page. The learner interactions are materialised through a scripting language and as a part of the MUPPLE page which stores one line of code per interaction. There learner interaction scripts,

of 50

then, can be shared manually, e.g. by sending an activity pattern to another user, or through an automated mechanism, i.e. recommendations of pre-defined PLEs (patterns) or single learner interactions within an activity (scripting code). In this version, MUPPLE has two important disadvantages. Firstly, it is implemented as a pure server-sided application which is problematic due to trust and privacy concerns and lacks the consideration of local applications. Secondly, learners require programming skills in order to de-personalize the learning experiences they want to share with others.

Schmitz et al. (2009) report on applying metadata, precisely the Contextualized Attention Metadata (CAM) set, for tracking the design and the usage of personal learning environments and recommending widgets to learners. A first realisation of this idea has been achieved with the Christmas Prototype in the ROLE project (cf. Renzel et al., 2010). Again, this approach includes the monitoring of user behaviour, whereby the pattern is captured in the form of metadata, and the sharing is restricted to an automated mechanism, the recommendation of tools. However, the control is with the user, allowing her to choose if the tracking data should be stored on the server or locally, or if tracking should be turned off. Storing interaction recordings locally can be considered a benefit here, as this data containing sensitive parts are stored at the user’s computer and can be shared as a pattern later on.

A.2 Survey of technology for activity pattern capturing The survey of pattern capturing technology, i.e. the formalisms behind a pattern sharing approach, is conducted along the six categories of the last section, namely (1) tracking and recording services, (2) personal and mash-up websites, (3) widget engines and learning management systems, (4) collaboration platforms, (5) workplace learning technologies, and (6) personal learning environment (PLE) solutions. Each of these categories provides some formalism to materialise learner interactions and realise user interaction recordings which then can be shared. In the following, relevant, prominent and useful solutions for pattern capturing are reviewed along these six categories.

Amongst the tracking and recording services, the Shareaholic browser plug-in simply uses the web-based API of the target platforms like Facebook, Delicious, Twitter, Google Mail, LinkedIn, etc. The materialisation is reduced to the required data fields to be submitted to these APIs, e.g. for adding a link to http://delicious.com a user has to specify an URL, a description and some optional data like tags or a specific date. Thus, the learner interactions are captured in the form of http requests storing specific data at the target platform. Regarding, the iMacros extension, the interaction recorder for the Firefox browser, materialises the recordings in form of files in the local directory of the browser add-on. Therefore, the providers of this add-on have developed a specific macro language with constructs like ‘TAB CLOSEALLOTHERS’ or ‘URL GOTO=http://www.iopus.com/imacros/demo/v6/extract2/’. Although not related to learning, these macros would work well for sharing learning experiences. However, they do not support de-personalization of sensitive data.

In the personal or mash-up websites category, Pageflakes as well as iGoogle seems to hide the digital recordings of the mash-ups designed and shared by users behind Javascript constructs or encrypted data. For instance, a pagecast (a user-given mash-

up of widgets) includes awkward and long values (e.g. ‘/wEPDwUKLTEzNzc3ODkxM[…]’, approximately 14.000 characters) in a hidden field named ‘__VIEWSTATE’. Yahoo Pipes allows modifying the tool mash-ups only by using a GUI-based mash-up editor.

Widget engines and widget-enabled LMSes, on the other hand, base their mash-up pages on html which, at the very best case, can be managed in a comfortable way through adequate drag-and-drop functionality. Elgg as well as the widget-enabled Moodle provide facilities so that end-users can specify URLs to (compliant) widgets. The state of the arrangement is stored in the underlying databases. Moreover, widget mash-ups provided by Wookie can be offered only through html pages or as part of widget engines like the Moodle plug-in. Also for special collaboration platforms like Google Wave the materialisation of activity patterns is not public or accessible at all. Reviewing the html code of a Wave indicates that the format is, at this early development stage, a proprietary json specification.

Finally, selected technologies for capturing activity patterns are reviewed in the field of workplace learning and personal learning environments. The following three approaches are examined closer: LISL, CAM, and PALADIN.

A first mechanism for capturing and sharing learner interactions has been developed in the scope of the iCamp project, below the surface of the MUPPLE prototype mentioned in the last subsection. Precisely, this platform was equipped with a domain-specific scripting language which was developed specifically for environment design. The so-called learner interaction scripting language (LISL) is used to (automatically) materialise the learner interactions with the environment while a user is designing her PLE and working with it. An example script for a MUPPLE page is given in Figure A-1.

Fig. A-1: Example LISL code

The figure shows that ‘define’ statements are used to create the underlying structure of the learning activity, whereby each activity consists of a set of learner interactions which are represented through action-outcome-tool triplets (lines 1-8). Connect statements allow learners to combine two tools with each other (line 9), requiring a certain degree of system interoperability of these web applications (cf. Wild, 2009). Action statements are used for starting the interaction by linking an action to an outcome and a tool and by launching the web-based tool specified for it.

of 50

of 50

The lines 10 to 12 of Figure A-1 show how three of these interactions (‘compose self-description using VideoWiki’, ‘browse peers using VideoWiki’ and ‘bookmark selected descriptions using VideoWiki’) are executed. In practice, the defined URL is opened in its own window on the mash-up space on the screen. Finally, line 13 gives an example of an interaction statement, here a drag-and-drop interaction, which is dependent on the visualization mode of the PLE solution. Define and action statements consist of user-given input, e.g. the name of the action or the URL, while interaction statements are pre-defined according to the possibilities of the PLE usage facilities.

When the web-based LISL interpreter executes such a script, the original MUPPLE page is restored to its last state including the model of the underlying activity and the tool mash-up visualisation. According to end-user development principles (cf. Repenning & Ioannidou, 2006), the MUPPLE prototype allows switching between the tool mash-up screen, a code editor for the LISL source of a page, and the LISL interpreter and its output. During pattern creation, LISL allows de-personalizing sensitive parts of the learner interactions. By using so-called placeholders, users can replace parts of the URL by a user-defined placeholder; for instance, line 2 of Figure A-1 contains the placeholder ‘%%peers%%’). After sharing such a pattern, the user who wants to re-instantiate the pattern as her own activity has to personalize it by entering a value for the placeholders.

The main criticism of the placeholder approach is that users have to de-personalize the activities manually in the code editor and thus must have LISL scripting skills, while all other functionality of MUPPLE can be achieved through web-based widgets and the LISL code is automatically generated in the background. Furthermore, a server-sided PLE solution like MUPPLE cannot track the usage of local applications, which might be of interest as evidenced with the Wakoopa approach. Finally, LISL is only focussing on the interactions of one learner, while collaborative activities have to be determined by synchronising the scripts of all peer learners.

The second formalism to materialise activity patterns is based on the CAM metadata schema. The CAM (Contextualized Attention Metadata) conceptual framework (see Wolpers et al., 2007) and the CAM schema enable the recording and management of rich and detailed sets of data about user attention, produced by any learning or non‐learning tool. The case studies developed have generated attention metadata from a variety of commonly used desktop applications, including PowerPoint, Winamp, MSN Messenger and Firefox.

The CAM schema (see Figure A-2) is designed to allow tracking user activities within any type of software that a user might use while working with a document (e.g. a web page, a text file, an image, an email, a MP3 music file, etc). The ‘feed’ element of the CAM schema is used to group the attention of the user in a certain software system. The ‘item’ element collects the attention given to a certain document. The ‘event’ element captures information related to every event in which the document was involved (e.g. reading, editing, updating, listening to, etc). The ‘action’ element provides information on the action that the document is involved in, while the ‘session’ element holds the information that is needed to identify the working session. Finally, the ‘context’

element captures information about the environments the user interacts with (e.g. information about a Moodle course the document has been downloaded from).

Fig. A-2: Overview of CAM metadata schema, taken from Wolpers et al. (2007)

The CAM framework enables the materialisation of learner interactions and activity patterns. However, the metadata schema does not consider aspects of de-personalizing the digital recordings so that learners cannot share their experiences without exhibiting sensitive data. Moreover, the pre-defined schema restricts the materialisation of patterns as well as the PLE infrastructure according to the main attributes of CAM, namely event, group, item, action, session, and context.

Thirdly, the PALADIN approach (‘PAttern LAnguage for DIsturbances in digital social Networks’, see Klamma et al., 2008) aims at combining digital social networks with a pattern-based automatic knowledge discovery technique, so that disturbances in social networks, like spammers sending irrelevant messages to a community, can be discovered and predicted. PALADIN analyses email communication, mailing lists, blogs, transaction-based websites, Wikis, URLs, and chat rooms and, consequently, materialises learner interactions on shared artefacts (e.g. messages, threads, blog entries, comments, web pages, feedback, etc.), in accordance with the Actor-Network Theory (Latour, 1999), in the form of networks (see example in Figure A-3) for which specific metrics (i.e. variables specifying roles like spammers, trolls, no answering persons, etc.) can be calculated.

of 50

Fig. A-3: PALADIN visualisation of a community with discovered patterns (conversationalists) in red

The user interactions of a networked community are captured in the form of a network with properties pre-defined for specific research questions. This approach is considered to be an effective formalism for analysing network effects and predicting behaviour of actors in this community. However, the PALADIN is not intended to share real user experiences but rather aims at detecting special characteristics in the network, like disturbances in social networks. Extending the possibilities of the before mentioned mechanisms and formats for capturing learner interactions, a pattern-based knowledge discovery technique like PALADIN can be useful to analyse and synchronise the user interaction recordings which might be distributed over different repositories or even on the computers of the learners.

A.3 Key features for activity pattern sharing The following table summarises all key features for activity pattern sharing identified in the state-of-the-art survey of the subsection A.1 and A.2. The list of features is structured along seven categories, namely user interactions, tools (widgets, applications), artefacts, actors, learning environment, activities, and pattern sharing.

Category Key feature Solution User interactions

Recording and replaying interactions with browser iMacros, Selenium

Editing iMacros, Selenium

Storing on disk or on server iMacros, Selenium

Sharing manually or via web service iMacros, Selenium, Shareaholic

of 50

of 50

Automated sharing Wakoopa, Audioscrobbler, RescueTime

Tagging and annotating interactions (user-given input) Shareaholic

Determining the user and usage context on a micro level

Shareaholic, APOSDLE, MUPPLE

Recommending events for the current work or scientific task

APOSDLE, AERCS

Materialising user interactions with a concrete formalism (macro or scripting language, JSON, etc.)

iMacros, Selenium, MUPPLE

Automated and manual materialisation of interactions

iMacros, Selenium, MUPPLE

Recording user attention for events CAM

Tools, widgets

Managing URLs locally or over various bookmarking sites Shareaholic

Tracking desktop and web applications Wakoopa, RescueTime

Recommending tools for specific situations

Wakoopa, RescueTime, iGoogle

Sharing tools manually iGoogle, MyYahoo

Including self-developed tools iGoogle, MyYahoo

Combining tools with each other Yahoo Pipes, MUPPLE

Providing appropriate tools for the current work task (situation) APOSDLE

Binding tool to user interaction (specifying the purpose of each tool) MUPPLE

Recording user attention on using a tool CAM

Artefacts Organising ones research library Mendeley

Sharing academic papers Mendeley

Sharing practices with others Mendeley

Discovering new materials and trends Mendeley

Collaborating on shared artefacts, recording changes and visualising the user contributions

Google Wave

of 50

Integrating web functionality i.e. widgets Google Wave

Replaying developmental history of shared document Google Wave

Adding and removing users at any time of collaboration Google Wave

Recommending documents for the current work task APOSDLE

Recording user attention given to a document CAM

Actors Recommending experts for the current work task or scientific activity

APOSDLE, AERCS

Analysing the network structures behind the collaborative activities (calculate specific metrics)

PALADIN

Identifying specific user roles PALADIN

Learning Environment To be designed by end-users

Pageflakes, Wookie-enabled LMSs, MUPPLE

Sharing with the community Pageflakes

Providing recommendations on the basis of tags and topicality Pageflakes

Integration of widgets Wookie-enabled LMSs

Integration of full-features web applications MUPPLE

Browsing and retrieving widgets available

Wookie-enabled LMSs

Determining the user and usage context on a macro level

APOSDLE, MUPPLE

Drag-and-drop functionality for designing PLEs

Wookie-enabled LMSs, MUPPLE

Giving the user the full control over sensitive data

iMacros, Selenium, CAM

Allowing to turn tracking on and off CAM

GUI-based mash-up editor Yahoo Pipes

Usage tracking according to pre-defined interactions given by the PLE solution

iMacros, MUPPLE

Allowing to switch between different visualisation modes, i.e. different

MUPPLE

of 50

facilities for PLE experts and novices

Recording and managing rich and detailed data sets about user attention CAM

Activities Creating and managing user interaction recordings for each activity LAMS, MUPPLE

Capturing user interactions automatically

LAMS, APOSDLE, MUPPLE

Managing user interactions manually MUPPLE

Designing activities by flat, hierarchical, and networked structures of user interactions

LAMS, APOSDLE, MUPPLE

Designing user interaction flows LAMS

Describing activities with metadata PLaNet, TLSDB, LAMS, etc.

Remove or replace user-specific data (de-/re-personalization)

Yahoo Pipes, MUPPLE

Pattern sharing

Motivating users to share e.g. by visualising their benefits in connection with privacy issues

APOSDLE

Supporting users in identifying sensitive data to be de-personalized -

Enabling users to de-personalize sensitive parts of their personal data

Yahoo Pipes, MUPPLE

Providing different pattern repositories pluggable into one’s PLE solution and allowing to share and retrieve good practices over a public API

Last.fm, Mendeley, LAMS

Reminding users to share good practices after completing an activity Facebook

Supporting information pull through manual retrieval of new patterns (browsing, searching)

iMacros, Selenium, Shareaholic, Mendeley, MUPPLE

Supporting information push through automated mechanisms like recommendations

APOSDLE, MUPPLE, AERCS, Facebook

of 50

References Alexander, C., Ishikawa, S., and Silverstein, M.: A pattern language: Towns, building,

construction. Oxford University Press, Oxford, 1977.

APSODLE Consortium: Third Prototype APOSDLE: Deliverables D1.9, D2.9, D3.9, D4.9, D5.9. APOSDLE project, 2009, http://www.aposdle.tugraz.at/media/multimedia/files/third_prototype_aposdle (2010-01-28).

Ben Youssef, A., and Dahmani, M.: The Impact of ICT on Student Performance in Higher Education: Direct Effects, Indirect Effects and Organizational Change. The Economics of E-learning, 5(1), 2008, pp. 45-56.

Clarke, R.: What’s Privacy? Proceedings of the Workshop at the Australian Law Reform Commission, 2006, http://www.rogerclarke.com/DV/Privacy.html (2010-02-09).

Dalziel, B.: Designing LAMS templates for medical education. Proceedings of the International LAMS Conference, Sydney, 2007, pp. 43-49.

Danish Technological Institute: Supporting Digital Literacy: Public Policies and Stakeholders’ Initiatives. Project description summary, 2007, http://www.digital-literacy.eu/_root/media/24823_Project%20description.pdf (2010-01-28).

Das, S., Egecioglu, Ö., and El Abbadi, A.: Anonymizing Edge-Weighted Social Network Graphs. UCSB Computer Science Technical Report 2009-03, 2009.

Derntl, M., Neumann, S., and Oberhuemer, P.: Report on the Standardized Description of Instructional Models. Deliverable 3.1, ICOPER project, 2009, http://dspace.ou.nl/dspace/bitstream/1820/2057/1/D3.1.pdf (2010-02-09).

Dillenbourg, P.: Designing Biases That Augment Socio-Cognitive Interactions. In R. Bromme, F.W. Hesse, and H. Spada (Eds.): Barriers and Biases in Computer-Mediated Knowledge Communication, Springer, New York, 2005, pp. 243-264.

Dwyer, C., Hiltz, S.R., and Passerini, K.: Trust and privacy concern within social networking sites: A comparison of Facebook and MySpace. Proceedings of Americas Conference on Information Systems, Keystone, 2007.

Eckstein, J., Manns, M.L., and Voelter, M.: Pedagogical Patterns: Capturing Best Practices in Teaching Object Technology. Software Focus, 2(1), Wiley, 2001.

Eder, J., Stark, K., and Zatloukal, K.: Achieving k-anonymity in DataMarts used for gene expressions exploitation. Journal of Integrative Bioinformatics, 1(4), 2007.

El Helou, S., Salzmann, C., Sire, S., and Gillet, D.: The 3A Contextual Ranking System: Simultaneously Recommending Actors, Assets, and Group Activities. Proceedings of the ACM Conference On Recommender Systems (RecSys), New York, 2009, pp. 373-376.

European Commission: Digital Literacy: Skills for the Information Society, 2007, http://ec.europa.eu/information_society/tl/edutra/skills/index_en.htm (2010-01-28).

European Parliament and the Council of Europe: Recommendation of the European Parliament and of the Council of 18 December 2006 on key competences for lifelong learning. Official Journal of the European Union, L394, 2006, http://eur-lex.europa.eu/ LexUriServ/site/en/oj/2006/l_394/l_39420061230en00100018.pdf (2010-01-28).

of 50

García-Barrios, V.M.: User-centric Privacy Framework: Integrating Legal, Technological and Human Aspects into User-Adapting Systems. Proceedings of the IEEE International Conference on Information Privacy, Security, Risk and Trust (PASSAT), Vancouver, 2009.

Henning, V., and Reichelt, J.: Mendeley - A Last.fm For Research? Proceedings of the IEEE International Conference on eScience, Indianapolis, 2008, pp. 327-328.

Henri, F., Charlier, B., and Limpens, F.: Understanding PLE as an Essential Component of the Learning Process. Proceedings of the World Conference on Educational Multimedia, Hypermedia and Telecommunications (ED-Media), Vienna, 2008, pp. 3766-3770.

Klamma, R., and Petrushyna, Z.: The Troll Under the Bridge: Data Management for Huge Web Science Mediabases. In H.-GT. Hegering, A. Lehmann, J. Ohlbach, and C. Scheideler (Eds.): Proceedings of the 38. Jahrestagung der Gesellschaft für Informatik e.V. (GI), die INFORMATIK 2008, Bonn: Köllen Druck+Verlag GmbH, 2008, pp. 923-928.

Klamma, R., Cuong, P.M., and Cao, Y.: You Never Walk Alone: Recommending Academic Events Based on Social Network Analysis. Proceedings of the International Conference on Complex Science (Complex), Shanghai, 2009, pp. 657-670.

Klamma, R., Spaniol, M., and Denev, D.: PALADIN: A Pattern Based Approach to Knowledge Discovery in Digital Social Networks. Proceedings of the International Conference on Knowledge Management (I-Know), Graz, 2006, pp. 457-464.

Kolfschoten, G., Lukosch, S., Verbraeck, A., Valentin, E., and de Vreede, G.-J.: Cognitive learning efficiency through the use of design patterns in teaching. Computers and Education, 54, 2010, pp. 652-660.

Kooken J., Ley T., and De Hoog R.: How Do People Learn at the Workplace? Investigating Four Workplace Learning Assumptions. In E. Duval, R. Klamma, and M. Wolpers (Eds.): Creating New Learning Experiences on a Global Scale, Springer, Heidelberg, 2007, pp. 158-171.

Latour, B.: On Recalling ANT. In J. Law and J. Hassard (Eds.): Actor-Network Theory and After, Oxford, 1999, pp. 15-25.

Law, E., and Nguyen-Ngoc, A.V.: Fostering Self-directed Learning with Social Software: Social Network Analysis and Content Analysis. Proceedings of the European Conference on Technology Enhanced Learning (EC-TEL), Maastricht, 2008, pp.

Lindstaedt S. N., Ley T., Scheir P., and Ulbrich A.: Applying Scruffy Methods to Enable Work-integrated Learning. Upgrade: The European Journal of the Informatics Professional, 9(3), 2008, pp. 44-50.

Mayer, R.C., Davis, J.H., and Schoorman, F.D.: An Integrative Model of Organizational Trust. The Academy of Management Review, 20(3), 1995, pp. 709-734.

McAndrew, P., Goodyear, P., and Dalziel, J.: Patterns, designs and activities: unifying descriptions of learning structures. International Journal of Learning Technology, 2(2-3), 2006, pp. 216-242.

Motschnig-Pitrik, R., and Derntl, M.: Learning Process Models as Mediators between Didactical Practice and Web Support. Proceedings of the International Conference on Conceptual Modeling, Klagenfurt, 2005, pp. 112-127.

Motschnig-Pitrik, R., and Figl, K.: Researching the Development of Team Competencies in Computer Science Courses. Proceedings of the ASEE/IEEE Frontiers in Education Conference, Saratoga, 2008.

of 50

Mödritscher, F., and Wild, F.: Why not Empower Knowledge Workers and Lifelong Learners to Develop their own Environments? Proceedings of the International Conference on Knowledge Management (I-Know), Graz, 2009, pp. 268-277.

Mödritscher, F., Hoffmann, R., and Klieber, W.: Integration and Semantic Enrichment of Explicit Knowledge through a Multimedia, Multi-source, Metadata-based Knowledge Artefact Repository. Proceedings of the International Conference on Knowledge Management (I-Know), Graz, 2007, pp. 365-372.

Najjar, J., Duval, E., Ternier, S., and Neven, F.: Towards interoperable learning object repositories: the Ariadne experience. Proceedings of the IADIS International Conference WWW/Internet, Algarve, 2003, pp. 219-226.

Nguyen-Ngoc, A.V., and Law, E.L.: Perceived Usability of Social Software Enabling Self-Directed Learning. Proceedings of the World Conference on Educational Multimedia, Hypermedia and Telecommunications (ED-MEDIA), Toronto, 2008, pp. 1449-1458.

O’Reilly, T.: What is Web 2.0: Design Patterns and Business Models for the Next Generation of Software. O’Reilly Media Inc., 2005, http://oreilly.com/web2/archive/what-is-web-20.html (2010-01-29).

Persico, D., Pozzi, F., and Sarti, L.: Design patterns for monitoring and evaluating CSCL processes. Computers in Human Behavior, 25, 2009, pp. 1021-27.

Pfitzmann, A., and Hansen, M.: Anonymity, Unlinkability, Undetectability, Unobservability, Pseudonymity, and Identity Management: a consolidated proposal for terminology. Report, 2008, http://dud.inf.tu-dresden.de/literatur/Anon_Terminology_v0.31.pdf (2010-01-29).

Renzel, D., Höbelt, C., Dahrendorf, D., Friedrich, M., Mödritscher, F., Verbert, K., Govaerts, S., Palmér, M., and Bogdanov, E.: Collaborative Development of a PLE for Language Learning. International Journal of Emerging Technologies in Learning (iJET), 5, Special Issue on MashUps for Learning, 2010, pp. 31-40.

Renzel, D. Klamma, R., Spaniol, M.: MobSOS - A Testbed for Mobile Multimedia Community Services. Proceedings of the International Workshop on Image Analysis for Multimedia Interactive Services (WIAMIS), Klagenfurt, 2008, pp. 139-142.

Resnick, P., and Varian, H.R.: Recommender systems. CACM, 40(3), 1997, pp. 56-58.

Schmitz, H.-C., Scheffel, M., Friedrich, M., Jahn, M., Niemann, K., Wolpers, M.: CAMera for PLE. Proceedings of the European Conference on Technology Enhanced Learning (EC-TEL), Nice, 2009, pp. 507-520.

Shaffer, D.W., Hatfield, D., Svarovsky, G.N., Nash, P., Nulty, A., Bagley, E., Frank, K., Rupp, A.A., Mislevy, R.: Epistemic network analysis: A prototype for 21st century assessment of learning. International Journal of Learning and Media, 1(2), 2009, pp. 33-53.

Sobernig, S., Danielewska-Tulecka, A., Wild, F., and Kusiak, J.: Interoperability and Patterns in Technology-Enhanced Learning. XVIII Górska Szkoła Polskiego Towarzystwa Informatycznego w Szczyrku, Polish Information Processing Society (PTI), Szczyrk, 2006.

Stucky, W., Dixon, M., Bumann, P., and Oberweis, A.: Information Technology Practitioner Skills in Europe: Current Status and Challenges for the Future. In R. Klein, H.-W. Six, and L. Wegner (Eds.): Computer Science In Perspective, Springer, Berlin, 2003, pp. 304-317.

Thoman, E., and Jolls, T.: Media literacy education: Lessons from the center for media literacy. In G. Schwartz, and P.U. Brown (Eds.): Media literacy: Transforming curriculum and teaching, National Society for the Study of Education, Malden, 2005, pp. 180-205.

of 50

Van Harmelen, M.: Personal Learning Environments. Proceedings of the IEEE International Conference on Advanced Learning Technologies (ICALT), 2006, pp. 815-816.

Van Harmelen, M.: Design trajectories: four experiments in PLE implementation. Interactive Learning Environments, 16(1), 2008, pp. 35-46.

Wild, F. (Ed.): Mash-Up Personal Learning Environments. Deliverable D3.4, iCamp project, 2009, http://www.icamp.eu/wp-content/uploads/2009/01/d34_icamp_final.pdf (2010-01-29).

Wild, J., Wild, F., Kalz, M., Specht, M., and Hofer, M.: The MUPPLE competence continuum. Proceedings of MUPPLE Workshop at the European Conference on Technology Enhanced Learning (EC-TEL), Nice, 2009, pp. 80-88.

Windschitl, M., and Sahl, K.: Tracing teachers’ use of technology in a laptop computer school: The interplay of teacher beliefs, social dynamics, and institutional culture. American Educational Research Journal, 39(1), 2002, pp. 165-205.

Wolpers, M., Najjar, J., Verbert, K., and Duval, E.: Tracking Actual Usage: the Attention Metadata Approach. Educational Technology & Society, 10(3), 2007, pp. 106-121.

Zhou, B., Pei, J., and Luk, W.: A brief survey on anonymization techniques for privacy preserving publishing of social network data. ACM SIGKDD Explorations Newsletter, 10(2), 2008, pp. 12-22.