Collaborative Metadata Management Claudio Jossen Department of Informatics University of Zurich Database Technology Research Group Binzmühlestrasse 14, CH-8050 Zürich e-mail: jossen@ifi.unizh.ch, http://www.ifi.unizh.ch Tel.: +41-44-635 6753, Fax: +41-44-635 6809
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 2
Motivation Joint work with the largest Swiss Health Insurance Company: Helsana Ca. 2 Million Customers New Data Warehouse Introducing SOA / SCA Well-established Processmanagement Problem: No central and standardized Metadata Interface (API / GUI) 3
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 4
Kinds of Metadata 5
Metadata Management I 6
Metadata Management II Source: : Meta Data Repository Redux, Part 1 & 2, DM Direct April 2004, www.dmreview.com 7
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 8
Metadata-Dimensions Enterprise Information Model Sources Model Subject Area Dimension 1: Da ata Hierarchy Model Subject Area Entity Attribute Model Subject Area Entity Data Warehouse Model Model Subject Area Subject Area Entity Entity Attribute Attribute Dimension 3: Abstraction Reporting Model Subject Area Entity Attribute Entity Attribute Dimension 2: Data Flow 9
Metamodel Data schema <rdf:description rdf:about=http://www.helsana.ch/mdm/models/bdwh> <rdf:li> <rdf:description rdf:about=http://www.helsana.ch/mdm/entities/adr> <rdf:li> <rdf:description rdf:about=http://www.helsana.ch/mdm/attributes/adr/sprache_c> <mdmerwin:parentattribute rdf:resource=http://www.helsana.ch/mdm/attributes/adr_typ/sprache_c/> <mdmerwin:code rdf:resource=http://www.helsana.ch/mdm/code_templates/bdwh/sprache_c/> </rdf:description> </rdf:li> </rdf:description> </rdf:li> </rdf:description> 10
Metamodel - ETL <rdf:description rdf:about=http://www.helsana.ch/mdm/transformations/sourcedb1001> <mdmetl:type>s</mdmetl:type> <mdmetl:step rdf:resource=http://www.helsana.ch/mdm/steps/sourcedb1001_1> </rdf:description> <rdf:description rdf:about=http://www.helsana.ch/mdm/steps/sourcedb1001_1> <mdmetl:source rdf:resource=http://www.helsana.ch/mdm/attributes/t_d H_RECH/sender_ean/> <mdmetl:source rdf:resource=http://www.helsana.ch/mdm/attributes/t_dh_rechpos/ean_responsible/> <mdmetl:target rdf:resource=http://www.helsana.ch/mdm/attributes/ean/ean/> </rdf:description> L: Lookup C: Case B: Calculation S: Standard G: Generated 11
Metamodel Information model <rdf:description rdf:about=http://www.helsana.ch/mdm/entities/infomodel/adresse> <mdminfo:definition>die Adresse eines Partners </mdminfo:definition> <rdfs:subclassof rdf:resource= http://www.helsana.ch/mdm/entities/infomodel/partner> </rdf:description> <rdf:description rdf:about=http://www.helsana.ch/mdm/entities/adr> <rdfs:type rdf:resource= http://www.helsana.ch/mdm/entities/infomodel/adresse> </rdf:description> 12
Metamodel - Overview Http://www.helsana.ch/mdm/models/ Http://www.helsana.ch/mdm/entities/ mdmerwin:file_name mdmerwin:file_name mdmerwin:model_mart_path Http://www.helsana.ch/mdm/attributes/ mdmerwin:entity mdmerwin:definition mdmerwin:note Http://www.helsana.ch/mdm/domains/ Http://www.helsana.ch/mdm/subject_areas/ mdmerwin:note_2 mdmerwin:note_3 mdmerwin:subject_area mdmerwin:definition mdmerwin:referenced_entities mdmerwin:udp mdmerwin:logical Only mdmerwin:domain mdmerwin:domain_name mdmerwin:type mdmerwin:datatype mdmerwin:logical_datatype 13 Http://www.helsana.ch/mdm/key_groups/ Http://www.helsana.ch/mdm/udp_definition_groups/ mdmerwin:logical_only mdmerwin:physical_name mdmerwin:comment mdmerwin:database mdmerwin:physical_only mdmerwin:index_generate mdmerwin:udp_row_count mdmerwin:type mdmerwin:view_with_check mdmerwin:view_where mdmerwin:view_group_by mdmerwin:view_having mdmerwin:view_order_by mdmerwin:view_sql mdmerwin:attribute mdmerwin:domain mdmerwin:logical_only mdmerwin:definition mdmerwin:note mdmerwin:udp mdmerwin:physical_name mdmerwin:datatype mdmerwin:not_null mdmerwin:physical_only mdmerwin:comment mdmerwin:physical_order mdmerwin:parent_attribute mdmerwin:physical_name mdmerwin:key_group_type mdmerwin:index_generate mdmerwin:index mdmerwin:key_group_member_column mdmerwin:udp_definition mdmerwin:type mdmerwin:definition mdmerwin:udp_list Http://www.helsana.ch/mdm/code_templates/ mdmerwin:database mdmerwin:query mdmerwin:template mdmerwin:code Http://www.helsana.ch/mdm/code_mappings/ mdmerwin:source mdmerwin:target
Architecture of the MDMS SeRQL / SPARQL SOAP HTTP Relational Database Web Service Container Java Web Container MDMS User SOAP Local Model Manager XMI files Data Modelling Application 14
Demo MDMS 15
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 16
Metadata Workflow 17
(Meta-) Model-Operators I µ 18
(Meta-) Model-Operators II στ 19
(Meta-) Model-Operators III µ: Transforms a simple RDF-Graph into a versioned and authoring related RDF-Graph σ: select - match - operator, returns an integrated RDF-Graph out of two input RDF- Graphs, where one of them has to be generated from the other by µ. τ: Reduces a versioned (temporal) RDF- Graph to a simple RDF-Graph 20
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 21
Collaborative Approaches Social Sciences: Communities of Practice (CoP) Computer Science: Web 2.0: Social Software Semantic Web: Semantic Social Software 22
Communities of Practice (CoP) intra-/inter-organizational community synchronous / ansynchronous distributed domain-based no timetable self-organizing 23
(Semantic) Social Software Forum -> thread-based Blog -> documentation Wiki -> versioned authoring Portal -> information/application integration And the winner is 24
(Semantic) Wiki 25
Agenda Motivation Metadata Management Metadata Management System (MDMS) Metadata Workflow Collaborative Metadata Management Conclusion 26
Conclusion The MDMS is now a stable and productive application with a growing number of users Implementation of the metadata authoring part still needs to be done In theory the wiki-approach fits the business requirements -> let s see what the users will do with it 27
Further Research Integration of Service-Metadata: RDF(S) und SOA (SAWSDL) Connection and Visualization of Business Processes (BPEL) Upgrade to a full Knowledge Management Portal (IT, Processes, Organization) 28
Questions? Comments? Wer sonst hat soviele Metadaten? 29