Presentation on theme: "Ballot Resolution on ISO/IEC DTR 20943-5 MMP Tae‐Sul Seo (KISTI*, Sung‐Joon Lim (KDB**, * Korea Institute of Science."— Presentation transcript:
Ballot Resolution on ISO/IEC DTR 20943-5 MMP Tae‐Sul Seo (KISTI*, firstname.lastname@example.org) Sung‐Joon Lim (KDB**, email@example.com) * Korea Institute of Science and Technology Information ** Korea Database Agency ISO/IEC JTC1/SC32/WG2 N1795
Contents Summary of 20943-5 Brief history of 20943-5 Summery of comments and discussion Acknowledgement
Summary of 20943-5 20943-5 describes a metadata mapping procedure (MMP) maximizing the interoperability among MDRs. MDR AMDR B
Type number Type Sub-Type number Sub-TypeMarkExamples Ways of harmonization (Types of mapping) 1Identical 1.1Identical One-to-one mapping 2Hierarchical 2.1GeneralizationH/genRetail price, Wholesale price Price 2.2SpecializationH/spePrice Retail price, Wholesale One-to-one mapping (dumb down) 2.3CompositionH/comFamily name, given name Name 2.4DecompositionH/decName Family name, given name One-to-many or many-to-one mapping (if required) 3Domain 3.1DomainDSummary : Synopsis One-to-one mapping (if required) 4Lexical 4.1SynonymsL/synFirst name : Given name 4.2AbbreviationL/abbAddress : Addr. 4.3AcronymsL/acrSerial Number :SN 4.4Case sensitivityL/casAddress : ADDRESS 4.5LanguageL/lan Name : 이름 4.6VariationL/varColor : Colour One-to-one mapping 5Syntactic 5.1OrderingS/ordFamily name : Name (family) 5.2DelimitersS/delFamily-name : Family_name 5.3MissingS/misAuthor name : Author One-to-one mapping 6Complicated 6.1ComplicatedC Mapping is impossible. Types of heterogeneity Summary of 20943-5
Summary of comments Comments from Canada(14) and ISO/CS(6) – All comments are editorial and were accepted and reflected. Comments from US(26) – Almost all comments were accepted and reflected except some comments required to be discussed.
Discussion US012 (including US013, 014 and 26) – Comment The procedure could be applied to DEC’s describing the data in statistical data sets. The elements do not have to be metadata elements. The metadata in the title of this Part refers to the descriptions of the DEC’s, not whether such DEC’s describe metadata. There, the title of the top box appears wrong. Yes, DC, MARC, and other specs are described as metadata standards. But, they still are schemas for some data. – Proposal Change “metadata element sets” to “sets of DEC’s”. Then, shorten “data element concepts” in the titles in the other 2 boxes to “DEC’s”.
Discussion US012 (including US013, 014 and 26) – Resolution It is just a request for changing expression. For the efficiency of editing work we insert DEC into the definition of “metadata element sets” instead of changing the words. Sub-clause 3.1.9. metadata element set small and fundamental group of data elements or data element concepts through which resources can be described and catalogued in a domain EXAMPLEDC(Dublin Core), MARC(MAchine Readable Cataloguing), MODS(Metadata Object Description Schema), etc.
Discussion US015 – Comment The CD and VD case is described in 4.2 and seems disconnected from 4.1. By addressing them together, we ensure data interoperability, since each datum is a designation of a VM, at its lowest level. For example, a datum described by the DEC “marital status of person” also identifies the status a particular person belongs: single, married, widowed, and divorced. So, data interoperability, the goal of harmonization procedures, requires looking at both DEC’s and CD’s together. – Proposal Tie sub-clauses 4.1 and 4.2 more tightly together. In particular, note how the CD is used to classify objects corresponding to the OC according to the Property in the DEC. Even though DEC’s can be the same, their CD’s may differ substantially. This has an impact on how well data may be translated from one system to another. This is why making a stop point at the DEC may not make sense.
Discussion US015 – Resolution To tie sub-clauses 4.1 and 4.2 more tightly, we changed the first sentence of 4.2 (now 4.3) to “For data element concepts which have been mapped, CD and VD must be … harmonized as well.”
Discussion US016 – Comment To evaluate DEC’s, the order requires looking at OC’s then P’s. However, one may have the following 2 DEC’s: Sex of bears and Sex of deer Even though the OC’s differ, the P’s are the same. It may be possible to combine the data described in some meaningful ways, and if we look at the OC first, we may miss the similarity exposed through the use of the same P. – Proposal The procedure should take this into account. In comparing DEC’s, it makes sense to organize by P’s first in some circumstances.
Discussion US016 – Resolution If two or more object classes are related, the mapping procedure will be repeated as many times as the number of object classes. In case of the example, the mapping procedure will be performed twice for two object classes: bear and deer.
Discussion US017 – Comment Table 3 applies just as well to comparing VM’s as it does DEC’s, OC’s, and P’s. – Proposal Consistent with US-015 and US-016, incorporate this into an expanded method of comparing VM’s, P’s, and OC’s together. – Resolution Table 5 is just examples of table 3. The same principles are applied to CD and VD in this TR.
Discussion US018 – Comment The sub-clause mixes the ideas of comparing CD’s (through VM’s) with comparing VD’s (through designations, i.e., PV’s). The semantics are not contained in the PV in general. Some designations are terms, and therefore they are linguistic and convey meaning, but codes do not. Plus, you are missing the semantic comparisons available in each VM. – Proposal If you want to include a comparison of the designations in a VD, then add a new sub-clause to do this. Separate the VM comparison for CD’s from the PV comparison for VD’s. They aren’t the same. They don’t follow the same rules, and the rules that apply for VM’s are the same as for DEC’s, OC’s, and P’s.
Discussion US018 – Resolution Sub-clause 4.2 (now 4.3) already includes all rules required in comparison for CD and VD. Remember the title of this TR has been changed from SMMP to MMP.
Acknowledgement The editors of 20943-5 thanks to: – All WG2 members, – the conveners, and – Especially US and Canadian delegates who gave us many helpful comments.