Looking into the future Providing Social Science Data ServicesJim Jacobs
First principlesMetadata are data about data -- information about information.Its all about having complete, accurate, re-usable metadata.Software to process the metadata is secondary. We should be able to have metadata today that we know will be usable in unforeseeable computing environments (operating systems, software, hardware).
First principlesMetadata should be
ComprehensiveCompleteUncompromisedConsistentFlexibleSharableUsable and re-usablePreservableParseable by computerDocumentedNon-proprietary
How XML fits inXML is designed to be parseable with generic tools.XML can encode meaning and can be self-documentingXML is non-proprietary, open, flexible.
How XML fits inXML is designed to make it easy to find and use just the elements you need from a large document. Cherry picking
How XML fits in Great Power Wars, 1495-1815 9955 Levy, Jack S. National Science Foundation. SES86-10567 Inter-university Consortium for Political and Social Research 1994-05-20
1994-05-20 1994-05-20 Levy, Jack S. GREAT POWER WARS, 1495-1815 [Computer file]. New Brunswick, NJ and Houston, TX: Jack S. Levy and T. Clifton Morgan Great Power Wars, 1495-1815You can cherry- pick just what you need from a large XML document
From legacies to the futureSASSPSSOSIRISPDFPaperData dictionaryEtc.HTMLPDFAny stat packageNesstar, SDA, DataverseLibrary OPACGoogleOAI, METS, etc.RSS, RDFGISDDI 3, 4DDI
From many contributors to many usesresearcherData collectorAnalystData producer, distributorData archivistData librarianUsers of statisticsGovernment agencyThe webLive documentsDatabasespublicationsData archivesData librariesInstitutional repositoriesSecondary analysisNew researchNew knowledgeDDI
OAIS Functional ModelIngestOAIS Functional ModelArchival StorageAccess
Information PackagesSIPOAIS Information ModelAIPDIPSIPSIPDIPDIP
Data stewardship life cycle
DDI will enable transformationNew kinds of data discovery (beyond indexing)Metadata as a primary resource (metadata as data)
Metadata for data discoveryICPSR already uses DDI metadata to create its Variables database.Nesstar and Dataverse software use metadata to produce searchable indexes of data repositoriesIn the future we should see the harvesting of DDI from many repositories to create indexes across collections. (oclc.org/oaister/)In the future well see data discovery by concept and methodology and geography and time period, not just keyword.
Metadata as dataBy structuring metadata according to a methodology (the lifecycle-of-data approach), we create metadata that we can treat as data.We can analyze metadata the way we would analyze any data file.As more metadata of this kind are created, we are accumulating a body of information that makes it possible to study trends across time and geography.
Metadata as dataThe technical documentation for the Army's Korean conflict casualty electronic records file has casualty codes that were never used in the data files. The presence of codes in the metadata for injury by lethal gas and by radiation exposure suggests that Army personnel who designed this record-keeping system expected the possible use of those as weapons. Examination of the data alone would have missed this suggestion. The codes for 'place of casualty' included, in addition to South Korea Sector and North Korea Sector, the Indo-China Sector, Tibet Sector, Mongolia Sector, Honan Sector (sic), Manchuria Sector, North Japan Sector, South Japan Sector, South China Sector, and Formosa Sector."
Metadata as dataA researcher at the Danish Data Archive is doing a qualitative analysis of the questionnaires used in seven surveys about ethnic minorities in Danish society, "with the purpose of showing how surveys ... mirror and project societal understandings of the subjects under investigation."
Metadata as dataWendy Thomas of the Minnesota Population Center examined U.S. Census metadata from 1790 through 2000 and compared the changing concept of race and ethnicity as embodied in the categories used by the Census Bureau questions over time. Those concepts are only documented in the metadata, not the Census data files themselves.
******This means you can have a very large, very complex, very rich metadata file, pick out just the elements you need for a particular application and make those elements fit the requirements of the software.***So the functional model has ingest, storage, and access functions.***DDI will most likely be produced at the data production stage and the data repository stage of the life cycle. But contributions to ddi may be made at other stages as well.*But we will make use of ddi at every stage in the life cycle