Update for ISO/IEC PDTR Semantic Metadata Mapping Procedure (SMMP) November, 2010 Tae-Sul Seo and Sung-Joon Lim 1 DECCD DEVD (1:N) (N:1) MARC DCONIX GILS … Crosswalk ISO/IEC MDR ISO/IEC JTC1/SC32 WG2 Meeting California, USA WG2_N1474

Contents Brief history Comment resolution Working Draft – Scope – Normative references – Related standards – Semantic Metadata Mapping Procedure : Data Element – Semantic Metadata Mapping Procedure : Value Domain – Annexes Annex A.Types of Semantic Heterogeneities Annex B.Relationship to ISO MDR Future work 2

Brief history ~ /Study Period – Semantic Harmonization of Metadata /WG2 Resolution – SC32N1812/WG2N1215 Project sub-division was approved. – Semantic Metadata Mapping Procedure (SMMP) – Project Number /Summary of Voting – SC32N1856 Proposed SC32 resolution for sub-division was adopted /SC32 Resolution Draft – SC32N1887 Registration as PDTR was instructed ~/Update WD – WG2N1317 ISO/IEC PDTR – WG2N1392 ISO/IEC PDTR – WG2N1416 ISO/IEC PDTR /PDTR Ballot Request 3

Comment resolution Comments at the last meeting in Kunming Proposed change Enhance the example of value domain mapping add conceptual domain ResponseNew table for the example was addedConceptual domain was mentioned on the NOTE Resolution Mapping TypesExamples Conversion Code conversion KOR to KR Letter conversion KATS to Korea Agency for Technology and Standards Unit conversion 1USD to 1KRWON Structural rearrangem ent Compositio n First Name: John Given Name: Kennedy toto Name: John Kennedy Decomposit ion Name: John Kennedy toto First Name: John Given Name: Kennedy Order change John Kennedy to Kennedy, John NOTEThe conversion or rearrangement of the value is applicable only when the conceptual domains which govern the value domains to be mapped are same. If the conceptual domains are different to each other, value mapping may be performed partially or impossible. 4

Scope Background – Semantic metadata mapping is required to mediate among data elements to be interoperably used. – Metadata crosswalk needs to be elaborated to have semantics. Purpose – To set up a procedure for improving the semantic harmonization of metadata. Scope – The procedure is recommended for use in a specific domain. – The procedure considers the interoperability of data element and value domain. 5

Normative references ISO/IEC :2004 – Borrowing the fundamental model of data element. ISO/IEC :2005 – Borrowing the principles of naming convention and example in Annex A. CONCEPTUAL DOMAIN VALUE MEANINGPERMISSIBLE VALUES VALUE DOMAIN (1:N)(1:1)(N:1) 6

Related standards ISO WD – Interoperability with other vocabularies. – ISO 2788: ISO 5964:1985 = ISO 2596 – Types of heterogeneity (Hierarchical and lexical differences) are related. ISO/IEC WD – 5.2 Principles and rules for the development of a definition for a concept and assignment of an appropriate term in a controlled vocabulary – A kind of general rule for the development of term definition ISO/IEC – Language Accessibility and Human Interface Equivalencies (HIEs) in e-Learning Applications: Principles, Rules and Metadata Elements – A type of heterogeneity (domain difference) is related. 7

Semantic Metadata Mapping Procedure : Data Element Grouping data elements 1. Identifying metadata element sets to be mapped 2. Finding object classes 3. Grouping data elements by objects 4. Finding properties 5. Grouping data elements by properties 6. Finding data element concepts 7. Mapping by data element concepts 8. Recommending data element names 9. Giving notes about heterogeneity type Identifying metadata element sets Semantic Mapping Main ProcessesSub-Processes 8

Semantic Metadata Mapping Procedure : Data Element Premises of the SMMP – Object class should be same for data element concept mapping. – Data elements are semantically identical if data element concepts are same. – If data element concept (= object class + property) is same, the data element is semantically same. P R OC P Object Class Data Element Concept Data Element 9

Semantic Metadata Mapping Procedure : Value Domain The value of each data element may be converted or rearranged to solve the inconsistency of data type as below, when the data types are not equal. NOTE The conversion or rearrangement of the value is applicable only when the conceptual domains which govern the value domains to be mapped are same. If the conceptual domains are different to each other, value mapping may be performed partially or impossible. ConversionCode Conversion Letter Conversion Unit Conversion Structural RearrangementComposition Decomposition Order Change 10

Semantic Metadata Mapping Procedure : Value Domain Mapping TypesExamples Conversion Code conversionKOR to KR Letter conversion KATS to Korea Agency for Technology and Standards Unit conversion1USD to 1KRWON Structural rearrangement Composition First Name: John Given Name: Kennedy toName: John Kennedy DecompositionName: John Kennedyto First Name: John Given Name: Kennedy Order changeJohn Kennedy to Kennedy, John 11

Annex A. Types of Semantic Heterogeneities Same Domain Difference Lexical Difference Syntactic Difference Complicated Differences Hierarchical Difference Generalization/Specialization Composition/Decomposition 1 to 1 X Synonyms, Abbreviation, Acronyms, Case sensitivity, Language, Variation Type Sub-Type Mapping Type Ordering, Delimiters, Missing 1 to m m to 1 1 to 1 12

Annex B. Relationship to ISO MDR Identifying Namespaces Finding Object Classes Grouping DEs by Obj. Classes Finding Properties Grouping DEs by Properties Finding DECs Mapping by DECs Recommending DEs Giving Notes Read only user Context Obj. Class Property DEC DE Submitter Obj. Class Property DEC DE ISO/IEC MDR Metadata Mapping Agent 13

Future work ISO/IEC TR is being reviewed and improved to go into the next stage. – The WD is completing. PDTR ballot is required as soon as possible. – We will update the WD and submit it until March Acknowledgement – We would like to acknowledge and thank the delegates of Canada, Baba Piprani, for his support and advice in establishing value mapping types. His paper, A Model for Semantic Equivalence Discovery for Harmonizing Master Data, was a good reference of this standard. 14