9 th Open Forum on Metadata Registries Harmonization of Terminology, Ontology and Metadata 20th – 22nd March, 2006, Kobe Japan. Day: 3 Slot No. P20 Name:Ian Cornwell Organization: Mott MacDonald Metadata Registry for Intelligent Transport Systems
9 th Open Forum for Metadata Registry, Kobe, 2006 Intelligent Transport Systems
9 th Open Forum for Metadata Registry, Kobe, 2006 Intelligent Transport Systems
9 th Open Forum for Metadata Registry, Kobe, 2006 Intelligent Transport Systems
9 th Open Forum for Metadata Registry, Kobe, 2006 Intelligent Transport Systems
9 th Open Forum for Metadata Registry, Kobe, 2006 A goal of Intelligent Transport Systems: seamless door-to-door services which needs: integration of open systems from different organisations Without a registry this goal will be achieved later and at greater cost, as various organisations slowly find out how to integrate fragments of the overall service. Registry is important to I.T.S.
9 th Open Forum for Metadata Registry, Kobe, 2006 Operational In discussion Research contribution Highways Agency ITS Metadata Registry ITS Community metadata registry community English Highways Agency UK Travel Information Community
9 th Open Forum for Metadata Registry, Kobe, 2006 ISO Process Structure Registry Structure UML XML Schema ITS Metadata Registry Foundation Express in UML Convert to UML (via MOF- based tool) & include
9 th Open Forum for Metadata Registry, Kobe, 2006
Submission Paths
9 th Open Forum for Metadata Registry, Kobe, 2006 14 major models with over 15,000 registered items 10 different submitting organisations 3 out of 14 submissions as XML Schema 7 out of 14 as XMI from different UML tools XMI versions can vary but can bridge via XSL Registry Population
9 th Open Forum for Metadata Registry, Kobe, 2006 Registry Top Level
9 th Open Forum for Metadata Registry, Kobe, 2006 Card/DraftRecorded QualifiedPreferred Registry Process Mapped ISO roles to existing bodies All status levels found to be useful Process drove up the quality of submissions Deeper refinement needed to achieve harmonisation
9 th Open Forum for Metadata Registry, Kobe, 2006 Bottom up Agree and build on common data types Top down ITS Architecture, indexing of subject matter & function Middle Out Core Components Harmonisation Tactics Dealing with multiple overlapping submissions
9 th Open Forum for Metadata Registry, Kobe, 2006 Harmonisation of overlapping concepts Rely on submitters changing submissions? Make attributes the unit of re-use? Tag common attributes across classes? One union class with options + context? Core Components!
9 th Open Forum for Metadata Registry, Kobe, 2006 Core Component Business Information Entity Specific business context Independent of business context UN/CEFACT ebXML Core Components
9 th Open Forum for Metadata Registry, Kobe, 2006 Core Components
9 th Open Forum for Metadata Registry, Kobe, 2006 Relate classes, attributes, associations
9 th Open Forum for Metadata Registry, Kobe, 2006 Derive Core Components from specific models Our “ Core Components ” are actually superset of concepts in specific models, in a common subject matter area. Process as objective as possible to avoid Core Components being yet another competing model. –Don ’ t add or “ fix ” except when justified by existing models.
9 th Open Forum for Metadata Registry, Kobe, 2006 Variety across systems Not strictly compliant with UN/CEFACT Core Components But using the basic idea, registry UML representation copes
9 th Open Forum for Metadata Registry, Kobe, 2006 Build Conceptual Schema first … On the way to ontology In one case we started with taxonomy
9 th Open Forum for Metadata Registry, Kobe, 2006 … add attribute detail …
9 th Open Forum for Metadata Registry, Kobe, 2006 … all built by considering mappings from existing models
9 th Open Forum for Metadata Registry, Kobe, 2006 Value of Core Components Makes the similarities & differences explicit Mappings process distinguishes justified design from flawed design Generates objective feedback to submitters Use understanding when building translators Use to identify candidates for recommendations (or “ preferred ” status), awarded in a specific business context. All the thinking exposed to future designers
9 th Open Forum for Metadata Registry, Kobe, 2006 Conclusions on Results UML/XMI has given a successful technical foundation –Keeping costs low through alignment with standard tools –Only 3 out of 14 submissions as XML Schema Harmonisation in a mature domain needs something more than published registry processes –Core Components analysis evolving as a technique to fill this gap
9 th Open Forum for Metadata Registry, Kobe,