170 likes | 275 Views
deep & distributed e-learning resources discovery. JISC/CETIS Conference, Nov 2004 Parallel Session 1: Repository Boon Low Digital Library Division Edinburgh University Library. Project Information. Short names: discovery+ or d+ ELF Technical Development project (1st Call) Objective:
E N D
deep & distributed e-learning resources discovery JISC/CETIS Conference, Nov 2004 Parallel Session 1: Repository Boon Low Digital Library Division Edinburgh University Library
Project Information • Short names: discovery+ or d+ • ELF Technical Development project (1st Call) • Objective: • piloting services and a toolkit for resource discovery among heterogeneous repositories • Partners: • University of Edinburgh: Library and Medical College • University of Oxford, University of Southampton • Edina, Intrallect, WebCT
RepositoriesResource collections + Metadata Metadata Creation Submit Archive Search Access Management (ATHENS) Expose Cross search Federation Embedding in learning environments Deliver Domain-level Mapping Related IMS Digital Repository Specification use cases: ‘search/expose’, ‘request/deliver’, ‘submit/store’
ELF Common Services Mapping Messaging Authentication Authorisation Resolver DRM Metadata service registry Logging Identifier Filing Workflow Search Service registry Mapping Presence Rules Harvesting E-mail management Scheduling Content management Packaging Archiving Rating / Annotation Terminology User preferences Chat Federated search Group Person Role Member Calendaring Metadata management Format conversion Alert Whiteboard Forum AV conferencing Context
interoperability service interface: SRU/OpenURL d+ services d+ services d+ services d+ toolkit metadata container: SRW/U to IMS CP, RLI d+ framework metadata: MARC, GRS, Proprietary to LOM, RLI, DC reference services consumer openURL Resolver Portal VLE ELF LD services Tools /Apps repository: z39.50, APIs, Web Services Library catalogues, ZETOC, RDN, ePrints UK, British Library, COPAC, Xgrain (ATHENS subscribed, e.g. EconLit, BIOSIS), COLIS (OCLC), ADLIB DSpace, IntraLibrary, PubMed, Google, Amazon, O’Reilly Safari, VLEs (WebCT, MVM) d+ system
Interoperability Demonstration • Service interface & repository interoperability • Federated search • Metadata/container mapping • Services integration in VLEs, Portals, Tools
deep & distributed e-learning resources discovery JISC/CETIS Conference, Nov 2004 Parallel Session 3: Repository Boon Low Digital Library Division Edinburgh University Library
interoperability service interface: SRU/OpenURL d+ services d+ services d+ services d+ toolkit metadata container: SRW/U to IMS CP, RLI d+ framework metadata: MARC, GRS, Proprietary to LOM, RLI, DC reference services consumer OpenURL Resolver Portal VLE ELF LD services Tools /Apps repository: z39.50, APIs, Web Services (CQL mapping) Library catalogues, ZETOC, RDN, ePrints UK, British Library, COPAC, Xgrain (ATHENS subscribed, e.g. EconLit, BIOSIS), COLIS (OCLC), ADLIB DSpace, IntraLibrary, PubMed, Google, Amazon, O’Reilly Safari, VLEs (WebCT, MVM) d+ system - recap
SRU/OpenURL Interface • Search & Retrieve URL (SRU) interface: unifying disparate repositories APIs and intefaces see: http://devil.lib.ed.ac.uk:8080/dplus/framework.jsp#facade • OpenURL for persistent linking • SRU uses Common Query Language (CQL) for specifying search strategy • query=title exact “grid computing” (title search) • query=“grid computing” and date exact 2004 (boolean) • CQL also provides context querying • Dublin core context, e.g. dc.title exact xxx • LOM context, e.g. lom.general.title exact xxx
SRU/OpenURL Interoperability • d+ OpenURL is based on SRU services • OpenURL query -> SRU query mapping • sid=?&atitle=?&author=? -> query= dc.title exact ? and dc.creator exact ? • Actions: • translate the URL in JSP/JSTL (string functions) • Delegate to the appropriate SRU service (c.f. ‘sid’) • SRU returns additional information/resource e.g. holding information and fulltext links, if the search returns unique results ( < 3 hits) • More elegant solution: a proper OpenURL/SRU translation serlvet
CQL LOM Context Sets (search fields) • CQL context sets (dc.title, lom.general.title) are used to specify search fields related to particular metadata schemas • c.f. other contexts (dc, bath), LOM CQL query context set can be long-winded e.g. “dc.creator exact ?” is equivalent to “ (lom.lifecycle.contribution.source exact LOMv.1.0) AND (lom.lifecycle.contribution.role exact AUTHOR) AND (lom.lifecycle.contribution.entry.string exact ?)” • Different repositories use different LOM context sets and profiles, lom.title (shortcut!), lom.author (shortcut!), lom.general.title or lom.general.title.string, lom.general.title.langstring.string - all valid in terms of SRW/U • Interoperability between different context sets is required • Action: lookup table via static ‘indexSynonym’ descriptor file
searchRetrieveResponse XML records recordData recordData recordData resource metadata resource metadata resource metadata Map to LOM, DC, RLI Metadata Interoperability • Mapping disparate native metadata to a specific schema is required so that the search results are consistent (for service consumption) • In addition to reusing native metadata schemas of the repositories, mapping services piloted: • MARC, GRS to LOM, DC, RLI • DC to LOM, RLI • Amazon, Google, Safari, Xgrain to LOM, DC
Metadata Interoperability • SRU accepts an optional ‘recordSchema’ parameter for selecting a target metadata schema, e.g. recordSchema=lom • Actions: • determine the target schema SRU ‘recordSchema’ parameter • determine the native or default (see next slide) schema • determine the mapping scenarios (if) • execute the mapping via XSLT service and the appropriate XML stylesheet
Metadata Interoperability • SRU does not specify how a default metadata schema should be selected from a sets of available schemas. • Most SRU targets provide a static default schema • LOC provides MARCXML (default), DC, MODS (2 variants), • RDN provides DC (default), IMS-LOM, IEEE-LOM • d+ is piloting a dynamic approach for implementing default metadata schema selection • Suggestion 1: “the default schema should be dynamically determined from the query context if the optional recordSchema is not specified” e.g.: • “dc.title” query gets DC records by default • “lom.title” query gets LOM records by default • “rli.title” query gets RLI records by default • “dc.title” query, with “recordSchema=LOM” parameter gets LOM records
searchRetrieveResponse XML IMS Content Package XML IMS Resource List XML records recordData recordData recordData resource metadata resource metadata resource metadata resources resourceList resource resource resourceMetadata metadata/lom resource resource resourceMetadata metadata/lom resource resource resourceMetadata metadata/lom Metadata Container Interoperability • mapping SRW/U to IMS Content Packaging and Resource List Interoperability Spec.
Metadata Container Interoperability • Action to map SRW/U to IMS CP/RLI: • Using the metadata mapping mechanisms, retrieve the appropriate SRU results containing the corresponding metadata schema of the container, I.e. CP->LOM, RLI ->RLI (metadata) • Map the SRU(LOM/RLI metadata) results with XSLT stylesheets, to IMSCP/LOM, or IMSRLI/RLI • Uses JSP & W3C XSLT mapping services as “post-search” process • More elegant action: • Modify SRW/U web services to return different containers in addition of the current “hard-wired” SRU response wrapper • Suggestion 2: “the separation of SRW/U service from its data binding” + suggestion 1 • e.g.: • “lom.title” query gets IMSCP/LOM records by default • “rli.title” query gets IMSRLI/RLI records by default • “dc.title” query gets SRW/DC records by default
further details • XSLT Styelsheets, Slide etc.: • http://devil.lib.ed.ac.uk/