collaborative management of business metadata

8/12/2019 Collaborative Management of Business Metadata

1/8

International Journal of Information Management 31 (2011) 366373

Contents lists available atScienceDirect

International Journal of Information Management

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / i j i n f o m g t

Collaborative management of business metadata

Kai M. Hner , Boris Otto, Hubert sterle

Institute of Information Management, University of St. Gallen, Mller-Friedberg-Strasse 8, 9000 St. Gallen, Switzerland

a r t i c l e i n f o

Article history:

Available online 17 January 2011

Keywords:

CollaborationData quality managementDesign science researchMetadata managementSemantic wiki

a b s t r a c t

Legal provisions, cross-company data exchange and intra-company reporting or planning proceduresrequire comprehensively, timely, unambiguously and understandably specified business objects (e.g.materials, customers,and suppliers). Onthe onehand, this business metadata has to cover miscellaneous

regional peculiarities in order to enable business activities anywhere in the world. On the other hand,data structures need to be standardized throughout the entire company in order to be able to performglobal spendanalysis, for example.In addition,businessobjects should adapt to newmarket conditionsorregulatory requirements as quickly and consistently as possible. Centrally organized corporate metadatamanagers (e.g. withina central IT department) arehardly able to meet allthese demands. They shouldbesupported by key users from several business divisions and regions, who contribute expert knowledge.However, despite the advantages regarding high metadata quality on a corporate level, a collaborativemetadata management approach of this kind has to ensure low effort for knowledge contributors asin most cases these regional or divisional experts do not benefit from metadata quality themselves.Therefore, thepaper at hand identifies requirements to be metby a business metadata repository, whichis a tool that can effectively support collaborative management of business metadata. In addition, thepaper presents the results of an evaluation of these requirements with business experts from variouscompanies and of scenario tests with a wiki-based prototype at the company Bayer CropScience AG.The evaluation shows two things: First, collaboration is a success factor when it comes to establishingeffective business metadata management and integrating metadatawith enterprisesystems, and second,semantic wikis are well suited to realizing business metadata repositories.

2010 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. Motivation and problem statement

A clear and unambiguous understanding of data is indispens-able for the effective management of multidivisional companies.At the same time, clear and unambiguous definitions of businessobjects (e.g. materials, customers, suppliers) are imperative forimplementing and benefiting from integrated, automated businessprocesses (Kagermann, sterle, & Jordan, 2010).Fundamentally,

smooth and efficient business process execution is ensured bymetadata describing the data structure of business objects (i.e.their attributes and relations) and providing information for cor-rect usage of business objects in business processes. High-qualitymetadata (i.e. metadata that is up to date, accurate and complete)helps to create and maintain a common understanding of businessobjects and business processes (Schmidt & Otto, 2008).It mini-

Corresponding author. Tel.: +41 71 224 2420; fax: +41 71 224 2777.E-mail addresses:[email protected](K.M. Hner),[email protected]

(B. Otto),[email protected](H. sterle).

mizes the occurrence of errors in automated activities and reduceswaiting times between activities. The following three examplesillustrate how poor metadata quality may negatively affect com-panies business process performance.

DB NetzAG (40,300employees, EUR4.1billionrevenuein 2009),a subsidiary of Deutsche Bahn AG,is responsible forGermanys rail-way infrastructure. The company has to make sure that its entireinfrastructure is accurately kept in an inventory database in orderto be able to report to national authorities on an annual basis forthe purposes of obtaining public funding. This task is impeded by

the fact that there is no company-wide definition of basic businessobjects (i.e. tracks, switches, and tunnels). For example, while atunnel in one geographical region is considered as an undergroundtrack segment, with the tunnel length being the distance betweentunnel entrance and tunnel exit, in another region tunnel lengthrefers to the distance between two underground stations. More-over, crucial knowledge (e.g. on the objects geographical position,maintenance, and monetary value) is distributed across the entirecompany. As a resultof these problems,any consolidation of infras-tructure data needed for inventorization is virtually impossible.

SBB Cargo AG (3700 employees, CHF 915 million revenue in2009), a subsidiary of Schweizerische Bundesbahnen, is responsi-

0268-4012/$ see front matter 2010 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ijinfomgt.2010.12.002
http://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ijinfomgt.2010.12.002http://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ijinfomgt.2010.12.002http://www.sciencedirect.com/science/journal/02684012http://www.elsevier.com/locate/ijinfomgtmailto:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ijinfomgt.2010.12.002http://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ijinfomgt.2010.12.002mailto:[email protected]:[email protected]:[email protected]://www.elsevier.com/locate/ijinfomgthttp://www.sciencedirect.com/science/journal/02684012http://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ijinfomgt.2010.12.002


2/8

K.M. Hner et al. / International Journal of Information Management31 (2011) 366373 367

ble for Switzerlands railway freight transportation. The trains ofSBB Cargo run on tracks that are used by a certain train type withina given periodof time. Tobook a track,SBB Cargo onlyneeds topro-vide thetrain type used andtheperiod of time. Nofurtherattributesare necessary to specify the business object from a data exchangeperspective. For internal purposes (e.g. specifying transportationorders, identifying dangerous goods to be transported, putting inplace work schedules for the staff), additional attributes are rele-vant, the values of which are set by various information systemsat various points in time. In order to be able to achieve flexibilityin planning and short response times (e.g. in case of disturbanceor malfunction), an unambiguous model of the business objects(e.g. track, train, train conductor) is needed that is understood ineverydepartment of the company.Another important requirementrelates to business object definitions of business partners, sinceduring any one trip the trains of SBB Cargo may run on tracks ofvarious providers in various countries, making it necessary to takeinto account different business object definitions when a track isbooked.

Bayer CropScience AG (18,700 employees, EUR 6.5 billion rev-enue in 2009),a marketleader in thefield of agrochemical products(e.g. for plant protection, insect control, seed growing), is a businessdivision of Bayer AG. Bayer CropScience is conducting a worldwide

business process harmonization project, in the course of whichmaster datais to be consolidated. Currently, the structure,meaningandusageof thecompanys master data vary from regionto region.For example, materials are counted by weight in Europe, and byitem (e.g. bottle, box) in other regions. In order to enable corporateanalyses (e.g. spend analysis, demand planning), data structuresand business metadata have to be harmonized as well.

1.2. Research objective and paper structure

A commonfeatureof the cases describedin Section 1.1 isthattheknowledge needed to create consistent business metadatais spreadacross various departments,divisions or lines of business. One pos-

sible solution to this problem could be to collaboratively identify,record, discuss and consolidate knowledge on various and differ-ing aspects of business objects (e.g. technical features, accountingattributes, regional characteristicsin thecase of DB Netzs tunnel)with the help of business experts. To do so, the use of a collabora-tion tool following the concept of Wikipedia1 (i.e. a software toolsupporting an iterative and collaborative approach for building arepository that any user can edit) could facilitate the process ofcollaboratively creating and maintaining consistentbusinessmeta-data.

A comparison of the cases described in Section 1.1 withWikipedia indicates the potential to use Wikipedia as a modelfor solving the problems described in the cases. Wikipedia evendescribes concepts like tunnel, train or train conductor,

and provides collaboratively maintained and commonly accepteddefinitions. If the maintenance of Wikipedia was restricted toemployees of DB Netz for example, the provided description oftunnel could be used as an accepted, company-wide definition.The question is, however, whether collaboration effects broughtabout by software similar to the Wikipedia platform can also beaccomplished in a business context. Therefore, the research objec-tives are (a) to identify business requirements to be met by arepository which supports the collaborative management of busi-ness metadata, (b) to implement a wiki-based business metadatarepository, and (c) to evaluate its use in a business context.

1

Wikipedias English website:http://en.wikipedia.org.

The next section of thepaperdefines business metadata accord-ing to literature on the subject. Furthermore, advantages anddisadvantages of wikis to be used for creating business meta-data repositories are discussed, and some scientific findings on theuse of wikis in business contexts are presented. Section3 speci-fies the research approach on which the paper is based and thetechniques used for identifying and evaluating requirements tobe met by a business metadata repository. Section 4deals withthe design of a business metadata repository, starting with theidentification of business requirements, followed by the deriva-tion of generic usage processes from these requirements, andfinally leading to the implementation of a wiki-based prototypethat meets the requirements and supports the usage processes.Section5 presents the results of a requirements evaluation withbusiness experts from various companies and users at Bayer Crop-Science using a wiki-based prototype. Finally, Section 6 presentsa summary of the paper and gives recommendations for futureresearch.

2. State of the art

2.1. Business metadata management

Metadata describes other data, i.e. it gives this data a certainmeaning in a certain context. For example, a stick has a length (aphysical property) of two meters (a measurement unit). Here, forthe description of an object (the stick) a data item (the number 2)and two metadata items (length and meters) are given ( Sen,2004,p. 151). Like the stick in this example, a business object isa real or virtual element in a companys process of providing ser-vices or products (e.g. a purchase or production order, a contract,a business partner, a certain material) that is characterized by anumberof properties represented by data elements (Martin, 1975).Metadata referring to a business object describe its characteristics,providing both technical aspects (e.g. data structure, data distri-bution) (Tozer, 1999,pp. 184194) and business-related aspects

like information regardingcorrect use of thebusiness object in cer-tain business processes (Burnett, Ng, & Park, 1999, p. 1213; Marco,2000, pp. 4952). Table 1 gives an overviewof the business objects,data attributes, and metadata mentioned in the examples givenin Section1.1,e.g. length of a tunnel in the case of DB Netz. Inthis context of metadata describing business objects, a businessmetadata repository is an information system that supports thecollaborative management of business metadata and that aims atdescribing business objects comprehensively, unambiguously andunderstandably (Schmidt & Otto, 2008, p. 215; Schreiber et al.,2000,pp. 1419).

2.2. Systems for collaborative metadata management

Dueto thesuccess of Wikipedia, companiesincreasinglyexpresstheir desire to use wikis for business purposes (Bughin, Chui, &Miller, 2008, p. 2, 2009, p. 14).As research has shown, the suc-cessful use of wikis in companies is basically possible (Majchrzak,Wagner, & Yates, 2006),mainly as effective tools for specificationprocesses (Wagner & Majchrzak, 2007). However, thedevelopmentof Wikipediahas shown thatunstructuredmetadata, as created andused in a classical wiki, may lead to redundancies and inefficientusage of wiki content (Krtzsch, Vrandecic, Vlkel, Haller, & Studer,2007).

In general, a wiki is a website providing content which can eas-ily be edited by any (registered or anonymous) user (Cunningham,1995).Depending on the software used, a wiki may come with dif-ferent functionality. There are two functions, however, that can

be found in practically any wiki (Kane & Fichman, 2009, p. 2):
http://en.wikipedia.org/http://en.wikipedia.org/http://en.wikipedia.org/


3/8

368 K.M. Hner et al. / International Journal of Information Management31 (2011) 366373

Table 1Examples of business objects, data attributes, and metadata.

Business objects Data attributes Metadata

DB Netz AG Rails, tunne ls, switche s, etc. Monetary value , geographical position,state of maintenance, etc.

Minimum requirements for specifications ofgeographical information, specification ofmeasures, differentiation of business objects (e.g.rail vs. switch), documentation of regionaldifferences, etc.

SBB Cargo AG Transportation orders, tracks, trains,

etc.

Track reservation period, track price,

train type, etc.

Documentation of different understandings of

business objects by different business partners,documentation of synonyms and homonyms (e.g.train conductor, train driver), etc.

Bayer Crop-Science AG Customers, suppliers, materials, etc. Name, customer address, materialdensity, material weight, packagingmeasures, etc.

Specification of unambiguous identifications ofbusiness partners, specifications for units ofmeasure, documentation of regional differences(e.g. address data, billing data), etc.

Wikipe dia Animals, companies, history, people ,technology, etc.

Categorization, denominators, date ofbirth, size, etc.

Documentation of historical facts, technicalexplanations, classifications of animals, people andplants, pictures, etc.

Open editing allows any user to change and modify content, andedit preservation serves for saving all edits and versions of thecontent. By means of a rollback, a previous version of certainwiki content can easily be reproduced (in order to prevent con-tent vandalism, for example) (Vigas, Wattenberg, & Dave, 2004,pp. 578580).

As mentioned above, metadata is largely unstructured in a clas-sical wiki. While the structuring of wiki content by means of designguidelines andthe linking of articles does support human users of awiki(cf. Wikipedia), it doesnot provideanyprerequisitesthat allowfor efficient machine processing of content (Krtzsch et al., 2007). Itis thisshortcoming that semanticwikisaddress, offering additionalfunctionality for content structuring including enhanced supportfor navigation and content search (Schaffert, Bry, Baumeister, &Kiesel, 2008).Semantic MediaWiki2 (SMW) is a software that pro-vides a semantic wiki (Krtzsch et al., 2007). It is an extensionof MediaWiki3 (the wiki engine of Wikipedia) and is capable ofprocessing annotated links between wiki articles as well as gener-ating a machine-processable representation of such relations. In a

semantic wiki that is used as a business metadata repository, forexample, links between wiki articles on a certain business object,its attributes, or process activities using the business object couldbe processed. Based on the machine-processable representation ofthe links, a list containing all attributes of the business objects or alist containing all business objects used in a certain process activ-ity could then be generated and used by other enterprise systemsautomatically. In a classical wiki, such lists need to be created andupdated manually.

An experiment comparing a classical wiki and a semantic wikiwith regard to usability and maintenance effort (Hner & Otto,2009) has brought about twobasicfindings: On theone hand, usersare able to find metadata more quickly, leading to higher satisfac-tion regardingusability of the system; on the other hand, the effort

formanagingandmaintainingthecontentofasemanticwikiiscon-sidered to be quite substantial since the information from whichthe machine-processable link structures are generated requiresadditional attention. As Section4 of this paper shows, however,enhanced usability is a success factor when it comes to fosteringthe willingness for collaboration across a company and collectingthe knowledge and information needed to produce high-qualitymetadata.

2 SMW website:http://www.semantic-mediawiki.org.3

MW website:http://www.mediawiki.org.

3. Research approach

The results presented in this paper have been developed in thecontext of theCompetence Center CorporateData Quality(CC CDQ).As part of the research program Business Engineering (sterle,1996), the CC CDQ is a consortium research project (sterle &Otto, 2010) aiming at the design of artifacts (e.g. architectures,methods, reference models) which are expected to solve practi-cal problems occurringin corporate data quality management. Theproject includes collaboration with subject matter experts fromvarious companies, who assist both in the process of problem iden-tification and in the process of evaluating the artifacts.

The research process applied follows the process modelfor Design Science Research proposed by Peffers, Tuunanen,Rothenberger, and Chatterjee (2008, pp. 5256). Design ScienceResearch is a research paradigm that aims at the design and eval-uation of solutions to practical business problems specified byorganizations (Hevner, March, Park, & Ram, 2004, p. 76; March &Smith, 1995, p. 260).Research in this context focuses on means-

end relations, not on cause-effect relations (Winter, 2008, p.470), e.g. a business metadata repository (means) to support busi-ness metadata management in a company (end).

This paper describesrequirements tobe metby a business meta-data repository as a Design Science Research construct (March &Smith, 1995, p. 256),the elements of which can be used for mod-eling further artifacts (e.g. generic processes for using the businessmetadata repository, software models). To validate the require-ments identified, a wiki-based prototype was implemented andevaluated in two ways: a focus group interview (Cavana, Delahaye,& Sekaran, 2001) was conducted to find out whether the prototypemeets the requirements and allows for effective metadata man-agement in organizations in general; and test scenarios at BayerCropScience were put in place to see if the prototype is more capa-

ble of supporting metadata management than a business metadatarepository already in use and considered inadequate.

4. Business metadata repository design

4.1. Requirements for effective business metadata management

The focus groups conducted in the course of the CC CDQdiscussed numerous cases where metadata of low quality (i.e.inaccurate, inconsistent and/or incomplete) had caused businessproblems (cf. Section1.1). It was agreed that business metadatamanagement should be supported by the collaboration of as manybusiness experts as possible from as many company divisions,

departments, and units as possible. The collaboration of distributed
http://www.semantic-mediawiki.org/http://www.mediawiki.org/http://www.mediawiki.org/http://www.semantic-mediawiki.org/


4/8


Fig. 1. Requirements to be met by a business metadata repository.

experts and metadata users was identified as a promising alter-native to centrally organized corporate metadata managers (e.g.withinacentralITdepartment)whoarehardlyabletoprovidecom-prehensive and timely information with regard to miscellaneousregional and divisional peculiarities.

Besides the positive effect of collaboration on metadata qual-ity (i.e. on comprehensiveness timeliness, understandability andunambiguity), the focus group also discussed the opposite causal-ity direction(cf. Fig. 1). Theparticipantsagreed thatthe highqualityof content as an outcome of a collaborative process might motivate

individuals to increase their contribution if they see the contri-bution of others as a payoff for their own effort. Two successfactors for achieving a high level of collaboration were identified:First, metadata should be easy to use for various business usersand easily accessible in as many enterprise systems as possible(e.g. customer relationship management); and second, the effortfor managing and maintaining metadata should be kept ratherlow.

The outcome of a focus group interview in 2007 (cf. AppendixAfor participants) is a list of requirements a business metadatarepository should meet (cf. Table 2) in order to enable collab-orative metadata management. Fig. 1 shows the interrelationsbetween therequirements to be metby a business metadata repos-itory and the two success factors for collaboration mentioned

above.

4.2. Processes for collaborative management of business

metadata

Bayer CropScience has been using a Lotus Notes-based busi-ness metadata repository. This repository contains metadata oncustomer, material and vendor attributes describing, among otherthings, theattributes maintenance in thecompanys ERP system. Inits current version, the repository is not used to the extent desired,leading to incomplete and outdated metadata. Table 3 lists theshortcomings of the business metadata repository used by BayerCrop-Science as identified in interviews with six users of the sys-tem. The interviews were conducted using a questionnaire (see

Appendix B)and were based on CommonKADS, which is a methodsupporting the development of knowledge intensive informationsystems (Schreiber et al., 2000, pp. 1419). The interview ques-tions are structured in three parts following a top-down approachto identify general tasks the interviewees use the repository for(questions A1A4), to identify usage processes for the tasks identi-fied (B1B8),and to address concretefunctionalityof therepository(user interface, different ways of using metadata, etc.) for each taskidentified (C1C4).

From the interview results, three generic usage processes for anew version of the business metadatarepository werederived. Oneof these processes (i.e. formaintaining metadata)is shown in Fig.2.Requirements to be met by a business metadata repository.

It addresses the requirements R10, R12R14, R17R20, R22 and

R23 (and thus also R01, R03 and R06R08), and describes the

reporting of a metadata defect by a metadata user and the valida-tion of metadata by a metadata manager. A metadata defect can bereported either by adding a comment to the respective metadataitem or simply by writing an e-mail to the respective metadatamanager. A metadata update is triggered either by a request for

Table 2Requirements to be met by a business metadata repository.

Requirement DescriptionR01. Cross-links

between metadataIn dictionaries, terms are usually defined with thehelp of cross-links to other terms, glossaries, orpictures. Likewise, the business metadatarepository should contain such cross-links in orderto avoid metadata redundancy and allow forefficient navigation.

R02. Standardizedmetadata format

The metadata kept in the business metadatarepository should be highly visible, i.e. it should beavailable and used by as many enterprise systemsas possible, thereby fostering the willingness totake part in the collaboration efforts among asmany users as possible. A standardized metadataformat would facilitate easy metadata import andexport.

R03. Flexible userinterface

The business metadata repositorys user interfaceshould offer options for flexible configuration (i.e.

different roles for different jobs) in order toaccomplish effective collaboration, including asmany users, departments and divisions as possible.

R04. Integration withportals and intranets

A frequent constraint stipulated by a companys ITstrategy is to reduce the amount of systems visibleto users and to provide a single front end (e.g. aportal or intranet). For that reason it should bepossible to integrate the business metadatarepositorys front end with existing informationsystems.

R05. Multilanguagecapability ofsoftware andmetadata

In order to accomplish effective collaboration (cf.R03), metadata must be available in differentlanguages. Also, the business metadata repositorysuser interface should be multilanguage capable.

R06. Support of usageand maintenanceprocesses

Usage and maintenance processes should beclearly specified and documented (e.g. searchingmetadata within the business metadata repository,workflows for updating metadata, or reportingdefects) in order to make sure the businessmetadata repository is efficiently used.

R07. Change history A metadata change history should be available atany time in order to be able to quickly identifymodifications made, correct errors, and recoverprevious states.

R08. Flexible metadatastructure

Over time, metadata is likely to requiremodification (regarding both meaning andstructure), caused by e.g. the need to describe newdata attributes or to implement additionalenterprise systems that use the data. In order toensure high metadata quality (i.e. up to date andaccurate), it should be easy to make modifications.

R09. Single-sign-onsupport

The business metadata repository should becapable of integrating existing modes forauthentication (i.e. no separate authenticationshould be required).


5/8


Table 3Functionality of a wiki-based business metadata repository.

Shortcoming Requirement Prototype functionality

No introduction anddocumentation

R10. Intuitive user interface (cf. R03, R06) Basic functionality from MediaWiki (e.g. full-text search) iscomplemented by automatically generated lists (SMWfunctionality). For clear navigation, metadata is clustered intothree groups (i.e. sorted by date of last quality check).

R11. User instructions and documentation in the wiki MediaWiki and SMW offer comprehensive user instructionsand documentation. Case-specific functions and configurations

have not yet been documented.Low quality of content R12. Feedback functions (cf. R06) Adding comments to articles is a standard feature ofMediaWiki.

R13. Easy identification of outdated metadata (cf. R06) Requirement is met by metadata being clustered into threegroups specifying the time they were last updated.

R14. Easy updating function (cf. R06, R07, R08) Data entry and editing is facilitated by special forms.Implementing a function for validation required an extension(about 40 lines of PHP code) in the case of Bayer CropScience.

No multilanguagecapability

R15. Multilanguage capability of user interface (cf. R05) The user interface of MediaWiki and SMW is available in anumber of languages.

R16. Support for structuring content available invarious languages (cf. R05)

The language the content is presented in can be chosen easily.Cross-links connecting different language versions of the samearticle have not yet been implemented.

Poor navigation R17. Cle ar navigation concept (cf. R03, R06) Navigation functionality of the prototype is designe d along useand management processes that are deduced from userinterview results.

Poorly structured searchresults

R18. Intuitive search function (cf. R06) The full-text search function provided by MediaWiki offers

state-of-the-art search functionality. Special filters or sortingmechanisms have not yet been implemented.

R19. Predefined filters (cf. R03) The classification of metadata into three groups specifying thetime they were last u pdated presents a simple predefinedfilter.

Unspecific denomination of metadata R20. Unambiguous denomination Metadata is clearly denominated by the article titles.No integration with ERP system R21. Integration with ERP system (cf. R02, R06) SMW is capable of exporting annotated content in various

formats (e.g. XML, CSV). Bayer CropScience is planning to takeadvantage of this option in order to be able to use themetadata of their wiki in their ERP system as well.

No documentation ofinterdependencies between fields

See R11R22. Support of content-related crosslinks (cf. R01) Cross-links between articles are a basic functionality of

MediaWiki.No user-specific views R23. Support of user-individual content and functions

(cf. R03)MediaWiki offers predefined wiki articles for registered users.Special templates for metadata managers are in preparation.

No cross-links between objects See R22

change (i.e.commentor e-mail) or by themetadatamanager check-ing a list of outdated metadata.

The other two processes cover the use of metadata directly inan enterprise system, and the joint use of enterprise systems anda business metadata repository for extended metadata search. Asthe requirements to be met by the specific repository used at BayerCropScience are largely identical with the general requirements arepository ofthis kind shouldmeet(cf. Table2), theusageprocesses

developedcan be seen as generic processesfor a business metadatarepository.

4.3. A wiki-based prototype

Taking into consideration the requirements and generic pro-cesses identified, a prototype was implemented and evaluated intwo scenarios (cf. Section5).This prototype is based on a semantic

Fig. 2. Process of maintaining business metadata.


6/8


Fig. 3. User interface of a business metadata repository.

wiki which was implemented by means of SMW (cf. Section2.2)

and a number of SMW extensions (cf. Semantic Forms4

for the cre-ation of forms for metadata maintenance).Table 3describes howthe prototypes functionality meets the requirements identified atBayer CropScience. In addition,Fig. 3shows some aspects of theprototypes user interface.

The business metadata repositorys main page (see Fig. 3a)presents a list of metadata sets in the order in which they weremost recently validated and updated by the respective metadatamanager (R13). Alternatively, metadata can be found using a full-text search.Fig. 3bshows parts of a business metadata repositorypage including information on who is responsible for the meta-data given. Each set of metadata is briefly described and the dateof the last validation is shown. By following content-related cross-links a user is able to identify the person responsible for the article

(R22). Validation dates may be colored, dependingon the time thathas passed since they were last updated (R13). A button labeledvalidate for indicating a content check is only displayed if theuser currently logged in is responsible for the metadata item given(R14).

The business metadata repository is configured so that only reg-istered users(i.e. metadatamanagers) areallowed to editmetadata,while comments can be added by any user (also anonymously).The respective manager is shown the metadata he is involvedwith or responsible for in predefined lists (R23). Fig. 3c showsa form for metadata maintenance that was created by meansof Semantic Forms. This extension is capable of analyzing themachine-processable wiki content in order to generate lists sup-porting data entry and editing, for example.

5. Artifact evaluation

5.1. Focus group evaluation results

Theprototypewasevaluatedbysubjectmatterexpertsinafocusgroup in 2008 (cf. Appendix Afor list of participants). In terms ofthe general requirements identified (cf. Table 2),this evaluationproduced the following results:

4 SemanticForms website:http://www.mediawiki.org/wiki/Extension:Semantic

Forms.

R01,R03,R07 and R08wereconsideredas beingmet bythe proto-

type. Consequently, these requirements were not discussed anyfurther.

R02 (standardized metadata format) was discussed very exten-sively. The possibility to integrate business metadata repositorycontent by means of metadata export (a standard functionoffered by SMW) was greatly appreciated. Bayer CropScienceis planning to use this function for integrating the busi-ness metadata repositorys content into their central ERPsystem.

R04 (easy integration with portals or intranets) was consideredas not being met by the prototype. However, it was agreed thatthis feature could easily be added, as the prototype has been setup as a standard website.

R05 (multilanguage capability) was considered as being met

by the prototype. Multi-language capability of the software isensured by MediaWiki and SMW both supporting a number oflanguages, with further languages being easily added, if needbe. Multilanguage capability of metadata can easily be achievedby any wiki, either by composing articles in various languagesor by linking articles written in different languages with oneanother.

R05 (support of usage and maintenance processes) was con-sidered as being partially met by the prototype through thepossibility to manage and maintain metadata by means of struc-tured forms (seeFig. 3c). Simple workflows can be supported bysequences of forms.

R09 (single-sign-on support) was considered as not being metby the prototype, as the prototype requires separate authenti-cation. However, it was agreed that the prototype could easilybe extended (e.g. extensions to integrate LDAP directories areavailable).

The subject matter experts agreed that in view of its intuitivefunctionality (cf. easy correction of errors and easy adding of com-ments) a wiki-based business metadata repository is very likelyto lead to increased collaboration. They also agreed that semanticwiki features are well suited for supporting non-technical businessusers (cf. forms for editing metadata and enhanced navigation forretrieving metadata). However, technical details of a semantic wiki(e.g. semantic annotations and semantic queries) should be hiddenfrom users in order to prevent an impediment to the positive col-laboration effect of using a wiki, especially in the case of business

users.
http://www.mediawiki.org/wiki/Extension:Semantic_Formshttp://www.mediawiki.org/wiki/Extension:Semantic_Forms


7/8


5.2. Results of prototype testing at Bayer CropScience

The prototypewas also evaluatedin the courseof functional testscenarios at Bayer Crop-Science. Following the generic processesdeveloped (cf. Section 4.2), a number of tasks were specified. Basedon the scenario shown inFig. 1.Requirements to be met by a busi-ness metadata repository, the test users, who had not been trainedto work with the prototype, were asked to

search for a certain wiki article, validate the metadata given in this article, report potential metadata defects to the persons responsible by

adding comments or by sending an e-mail, correct the metadata defects identified by the other test users, validate the modifications made.

Following the execution of the tasks specified, the test userswere asked in a final interview to

judge the prototypes usability, assess whether the prototype meets the identified requirements, give their opinion on whether more requirements to be met by a

business metadata repository could be identified.

All test users were able to execute all the tasks specified. Also,all test users confirmed that working with the prototype was noproblem for them and that the wiki offered very good assistancein performing the tasks successfully. As all test users were famil-iar with the companys regular business metadata repository, theywere able to assess the prototypes functionality in comparisonwith the old system. Aspects such as easy navigation, flexible andwell-structured lists, full-text search, and consistent cross-linkingof metadatawere considered as particularly positive. Only oneneg-ativestatementwasmadewithregardtothelowusabilityoflinkingexternal files to metadata in the repository.

6. Conclusion and future research

The paper identifies requirements to be metby a business meta-data repository for supporting the collaborative management ofbusiness metadata, specifies generic processes for such a repos-itory, describes the implementation of a wiki-based prototype,and presents the results of an evaluation of the requirements andthe prototype. The prototype analysis (see Table 3) shows thatmost requirements can be met by standard wiki functionality (e.g.feedback by means of comments, rollbacks, easy-to-use contentupdating, and cross-links) and that semantic wiki features are ableto meet remaining requirements (e.g. content structuring, meta-data export, and predefined content filters).

Further research should deal with the identification and devel-

opment of detailed functional specifications of a business metadatarepository, ensuring in particular metadata integration with enter-prise systems. In this respect, the possibility to export metadatain a standardized format could provide the functional basis forusing metadata directly in an enterprise system. Empirical studiesshould investigate whether using a business metadata repositorythat meets the requirements specified actually leads to the level ofcollaboration expected and desired (i.e. involving more or all of acompanys departments, divisions or lines of business in the pro-cess of collaborativebusinessmetadatamanagement)and whethermetadata quality can be significantly improved over time.

With regard to the case of Bayer CropScience presented in thispaper, further research should examine whether the scope of busi-ness metadata repository users can really be extended in the long

run, with more departments, divisions and lines of business of

Table 4Participants of focus group specifying requirements to be met by a business meta-data repository.

Company Function

alfabet AG Vice PresidentBayer CropScience AG AG Head of Master Data ServicesDai mler AG Proj ect Manager IT ManagementDB Netz AG He ad of Infrastructure Data ManagementDB Netz AG Head of Infrastructure Data Management Strategy

Deutsche Telekom AG Project Manager Master Data ManagementDeutsche Telekom AG Head of Data GovernanceE.ON AG Head of Master Data ManagementETA SA Project Manager Organization & ITETA SA Project Manager Organization & ITRobert Bosch GmbH Project Manager Master Data Managementsimple fact AG ConsultantZF Friedrichshafen AG Head of Organizational ServicesZF Friedrichshafen AG Project Manager Organizational Services

the company accepting and supporting the concept of collabora-tive business metadata management. Furthermore, it should beexamined whether a wiki as an open approach is capableof improving metadata quality substantially and with a sustainedeffect, or if editing rights shouldpossiblybe limited to a small num-

ber of people. Particular attention should be paid to the aspect ofintegrating metadata with the companys enterprise systems. Inview of the planned use of business metadata repositories in com-panies currently collaborating with the CC CDQ, it will be possibleto compare the analysis of similar cases in future research.

Appendix A. Focus group participants

OnNovember16,2007,subjectmatterexpertscollectedanddis-cussed requirements to be met by a business metadata repositoryduring a 45-min meeting as part of a two-day workshop. Table 4lists the functions of the participants in the respective companiesat the time of the workshop.

On December 3, 2008,subject matter experts identified and dis-

cussed strong points and weak points of the wiki described in thispaper during a 30-min meeting as part of a two-day workshop.Table 5 liststhe functionsof theparticipants in therespective com-panies at the time of the workshop.

Appendix B. Interview guideline for identifying use cases ofa business metadata repository

A1. Function and responsibilities at Bayer CropScience (depart-ment, business processes)?

A2. Introduction of the business metadata repository (training,documentation, informal activities)?

A3. Reaction and activitieswhen functionsof thebusiness meta-data repository are not understood?

A4. Tasks thebusiness metadata repository is used for (businessprocess, activity, information required)?

Table 5Participants of focus group evaluating the business metadata repository prototype.

Company Function

Alstom (Schweiz) AG Director Performance ImprovementB. Braun Melsungen AG Head of Central Material Master AgencyDB Netz AG Infrastructure Data Management TeamDB Netz AG Infrastructure Data Management StrategyDeutsche Telekom AG Data G overnance TeamRaiffeisen Schweiz Genossenschaft Quality Manager ITRaiffeisen Schweiz Genossenschaft Head of IT Risk ControllingSyngenta AG Global Supply IS ManagerTchibo GmbH Head of Master Data Management


8/8


B1. Actual business metadatarepository use (incidenttriggeringthe use of the handbook, search and navigation, metadata usage,going back to the task to be executed)?

B2. Metadata needed for a given task (business object, process,structure, guideline, rule)?

B3.Assessmentof metadata found (isit relevant,up to date, welldisplayed)?

B4. Frequency of business metadata repository usage?B5. Amount of (useful) metadata found? Possible reasons for

unsatisfactory search results?B6. Reaction and activities when desired information cannot be

found in the business metadata repository?B7. Weak points of the business metadata repository? Reasons

why it is notused more frequently (is metadata not displayed well,is it missing)?

B8. Suggestions for improvement (to increase usage of the busi-ness metadata repository)?

C1. Clarity of metadata structure? Is all the metadata neededquickly available?

C2.Whichoptionsfornavigationareusedandhowaretheyused(suggestions for improvement)?

C3. Which search options are used and how are they used (sug-gestions for improvement)?

C4. Which information tabs are used and how often are theyused (suggestions for improvement)?

References

Bughin, J., Chui, M., & Miller, A. (2008). Building the Web 2.0 Enterprise. McKinseyGlobal Survey Results.McKinsey Quarterly,2008, 110.

Bughin, J., Chui, M., & Miller, A. (2009). How companies are benefiting from Web2.0: McKinsey Global Survey Results. McKinsey Quarterly,2009, 1017.

Burnett, K., Ng, K. B., & Park, S. (1999). A comparison of the two traditions of meta-data development. Journal of the American Society for Information Science, 50 ,12091217.

Cavana,R., Delahaye,B. L.,& Sekaran,U. (2001).Applied business research: Qualitativeand quantitative methods. Milton, Australia: Wiley.

Cunningham, W. (1995). What is wiki. URL:http://www.wiki.org/wiki.cgi?WhatIsWiki, last accessed on 11. June 2010

Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information

systems research.Management Information Systems Quarterly,28, 75105.Hner, K. M., & Otto, B. (2009). The effect of using a semantic Wiki for metadata

management: A controlledexperiment.In R. H. Sprague (Ed.), 42ndHawaiiinter-national conference on system sciences. Hawaii: Waikoloa.

Kagermann, H., sterle, H., & Jordan, J. M. (2010). IT-driven business models: Globalcase studies in transformation. Hoboken, New Jersey: Wiley.

Kane,G. C.,& Fichman, R.G. (2009). Theshoemakers children:Using wikis forinfor-mation systems teaching, research, and publication. Management InformationSystems Quarterly,33, 117.

Krtzsch, M., Vrandecic, D., Vlkel, M., Haller, H., & Studer, R. (2007). SemanticWikipedia.Journal of Web Semantics,5, 251261.

Majchrzak, A., Wagner, C., & Yates, D. (2006). Corporate Wiki Users: Results of aSurvey.In D. Riehle (Ed.),2006 International Symposiumon Wikis 99-104. Odense,Denmark.

March, S.T., & Smith, G.F. (1995). Design andnaturalscienceresearch oninformationtechnology.Decision Support Systems,15, 251266.

Marco, D. (2000). Buildingand managing themeta datarepository.A fulllifecycle guide.New York: Wiley.

Martin, J. (1975). Computer data-base organization. Englewood Cliffs, New Jersey:Prentice-Hall.

sterle, H. (1996). Business engineering: Transition to the Networked enterprise.EM Electronic Markets,6, 1416.

sterle, H., & Otto, B. (2010). Consortium research: A method forresearcherpractitioner collaboration in design-oriented IS research. Business& Information Systems Engineering,2, 283293.

Peffers,K.,Tuunanen,T., Rothenberger,M. A.,& Chatterjee, S.(2008).A design scienceresearchmethodology forinformation systemsresearch.Journal of ManagementInformation Systems,24, 4577.

Schaffert,S., Bry,F., Baumeister, J., & Kiesel, M. (2008). Semantic wikis.IEEE Software,25, 811.

Schmidt, A., & Otto, B. (2008). A method for the identification and definition ofinformation objects. In P. Neely, L. Pipino, & J. P. Slone (Eds.), 13th Internationalconference on information quality(pp. 214228). Massachusetts: Cambridge.

Schreiber, G., Akkermans, H., Anjewierden, A., de Hoog, R., Shadbolt, N., de Velde,W. V., & Wielinga, B. (2000).Knowledge engineering and management: The Com-monKADS Methodology. Cambridge, Massachusetts: MIT Press.

Sen, A. (2004). Metadata management: Past, present and future. Decision SupportSystems,37, 151173.

Tozer, G. (1999).Metadata management for information control and business success.Boston, Massachusetts: Artech House.

Vigas, F. B., Wattenberg, M., & Dave, K. (2004). Studying cooperation and con-flict between authors with history flow visualizations. In E. Dykstra-Erickson,& M. Tscheligi (Eds.), Proceedings of the SIGCHI conference on human factors incomputing systemsVienna, Austria, (pp. 575582).

Wagner, C., & Majchrzak, A. (2007). Enabling customer-centricity using wikis andthe wiki way.Journal of Management Information Systems,23, 1743.

Winter,R. (2008). Design science researchin Europe. European Journal of InformationSystems,17, 470476.

Kai M. Hneris a research assistant and Ph.D. candidate at the Institute of Infor-mation Management at the University of St. Gallen. He is also a member of theCompetence Center CorporateData Quality andinvolvedin severalindustryprojectsconcerning corporate data quality management. His research interests includedesign of business oriented data quality metrics and development of software solu-tions for business metadata management. He holds a diploma in computer sciencefrom the University of Karlsruhe.

Boris Otto is senior lecturer and head of the Competence Center Corporate DataQuality (CC CDQ) at the Institute of Information Management at the University ofSt.Gallen.The CC CDQis a consortium research project aimingat designingartifacts

(e.g. architectures, methods, reference models, and software prototypes) for a sus-tainable quality management of corporatedata. He holdsa Ph.D. fromthe Universityof Stuttgart. Prior to hiscurrent position Boris Otto workedas a business consultantfor PricewaterhouseCoopers and SAP and as a senior researcher at the FraunhoferInstitute for Industrial Engineering.

Hubert sterle is founder and chair of the Institute of Information Managementat the University of St. Gallen and initiated the post-graduate course ExecutiveMBA in Business Engineering. His research interests include business models forthe information age, business networking, business engineering and independentliving. He holds a Ph.D. from the University of Erlangen-Nuremberg. Besides hisacademic career Hubert sterle founded The Information ManagementGroup (IMGAG) in 1989.
http://www.wiki.org/wiki.cgi?WhatIsWikihttp://www.wiki.org/wiki.cgi?WhatIsWiki

collaborative management of business metadata

Documents