F2F Pittsburgh Outcomes
From Library Linked Data
Use case discussion 9:00-10:30 + 11:00-12:30
Gathered Use Cases in clusters. See Use Cases.
Vocabularies 14:00-15:30 + 16:00-17:30
Main discussion topics for Vocabularies were as follows:
Vocabulary production & curation
- Ownership is important : FRBR core vs. FRBR-IFLA. Official owners should be quick in providing URIs, or else others will do it unofficially
- “Some institution” should maintain links between vocs /OR/ institutions who maintain vocs should provide links to other vocs // some political commitment is needed to improve trust in vocs, endorse them and curate them
- Persistence policy for ontologies and namespaces ?
- Guidance for creators of vocs : How to create good vocs for the Web ? Recipes, best practices, with a limited scope to library linked data.
- “Registering” vocabularies : i.e. declare them in a metadata registry with a URI. Is the term “registering” misleading ? The registry is not mandatory in a Linked Data perspective, but it brings some interesting additional services.
Use of vocabularies
- Perception of barriers regarding the use of certain vocabularies – certain more difficult than others, or perceived to be so
- Is it OK to pick some pieces from a voc, rather than following the global guidelines from the voc as it was originally intended ? “Out of band” guidelines are not significant in a LD environment, but they are in scope for our domain (libs). Vocs providers should be aware of risk of loss of context
- How to identify the source/scheme/context for controlled string values (vocabularies encoding schemes)
- MARC as the domain model for libraries ??? - Is a MARC ontology needed ? Or rather focus on content standards such as AACR2, ISBD, RDA ? Or MADS & MODS ?
- For the sake of Interoperability, need to adapt general vocs out there (e.g. FOAF) to library data : blank nodes ? Subproperties ? Contribute to development ? Anyway, seek convergence.
- Difference between the label and the concept – Authorities are about labels, names – not real world entities : the context is important !
- Current coverage of vocabularies – what's missing ?
Discussion on deliverables 17:30-18:30
Allocated curators to Use Cases clusters. See actions in the minutes.
Use cases & Topics 9:00-10:30 + 11:00-12:30
Grouped the Topics in 3 categories as follows:
Things that should be covered by use cases
- CM.CITATIONS (we have the Use Cases already)
- CM.NON-BIBLIO (to be pointed out : there is an AP for collections that includes administrative stuff. Don’t reinvent the wheel. Use external vocabularies.)
- LLD.SKOS-FOR-INTEGRATED-KOS. Use case needed to identify the requirement for a model more complex than SKOS (pending action on Gordon & Antoine).
- LLD.INSTITUTIONS-ID (Use case already available)
- SW.PROVENANCE, MGT.LICENSE, MGT.RIGHTS : to be extracted from existing use cases
- MGT.BIZ-MODELS: include ownership, return on investment and sustainability. To be extracted from existing use cases.
- DATA.SEARCH-OPTIMISATION: use case to be contributed by Antoine & Emmanuelle (pending action)
- DATA.VOCABS: need a use case for registries (RDA task group ?) - also includes : DATA.OBSOLETE-VOCABS, DATA.CHANGES, DATA-DISSEMINATION, DATA.VERSIONING, DATA.ONTOLOGY-DISCOVERY
- LLD.SKOS-EXTENSIONS: need to be captured in use cases. See the UC «Browsing & searching...» + contribute additionnal UC if needed. Check from SKOS postponed issues.
- Link Social Bibliography to a Bibliographic Network. Link reviews, tags, lists, cover art to a work. This seems like a variation on the enrich a record use case.
- Find stuff in several collections.Searching a specific set of library and non-library collections--for instance to find a given book or DVD, or get access to an article. I'm thinking about this first as a local area search with a specific set of libraries--but then also as a 'travel' awareness search (for the harder-to-find things I'm willing to go out of my way to find, while travelling). I'd love, for instance, a mashup between TripIt/Dopplr, my calendar of availability, and my LibraryThing wishlist -- alerting me to possibilities to use items I can't get locally when I have a free day out-of-town. Jodi Schneider 13:48, 30 June 2010 (UTC) It seems that we don't have any such far-stretching case. Or this item rather calling for enabling mesh-ups, not doing them? (Antoine)
- Find e-books see http://efoundations.typepad.com/efoundations/2010/07/finding-ebooks.html
- archives and linked data, knitting together multiple identifiers
Issues identified as requirements, and recommendations for future work
- Best practices should be issued regarding the modeling of FR’s vocabularies (CM.FRS). Ensure a better communication channel regarding the current status and updates on the development of domain vocabularies (LLD.VOCAB-STATUS)
- Provide explanation and advice on the reuse of legacy identifiers (ex. LCCN) as URLs (LLD.LEGACY-IDS)
- Clarify what an application profile is, and point to different approaches (documentation and guidelines vs. syntaxe for formal use) CM.APP-PROFILES. FR’s should be mentionned as a starting point (domain model in the Singapore Framework). Identify patterns for data models.
- Education and proof of concept regarding the shift from Record Paradigm to Graph Paradigm (SW.LD-VS-RECORD). Differentiate creation, dissemination and consumption of data: some of these could still need some form of record of aggregation. Record or aggregation packaging (SW.LD-PACKAGING).
- Need for coordination regarding the translation of data in MARC to Linked Data. Could it be done with an application profile ? (LLD.MARC2LLD, LLD.REFERENCE-MODEL-FIT)
- Identify different approaches regarding cross-linking of vocabularies (one-to-one vs. hub-spoke) SW.CROSS-LINKING-VOCAB
- Provide guidance on linked data project & management patterns (MGT. GUIDANCE, MGT. WORKFLOWS, MGT. PATTERNS). See the book «linked data patterns» and check how to adapt it to the case of libraries.
- Create a Library Linked Data Primer, including a glossary (MGT.TRAINING, MGT.LEGACY-MAPPING)
- Create formal and durable liaison with relevant standard bodies (MGT.STANDARDS-PARTICIPATION)
Non-library specific, but still relevant :
- Provide best practices on Namespaces (LLD.NAMESPACES), presistent identifiers resolution services, identifiers (SW.IDENTIFIERS), URI patterns (SW.REST)
- Provide best practices regarding Linked Data Architecture and Infrastructure, data caching, reference software. (DATA. ARCHITECTURE, DATA.CACHING)
Deliverables that the XG should achieve within its lifetime
- Framework for providing an inventory of available library linked datasets (related with CKAN) (LLD.AVAILABLE-LINKED-DATA). Inventory of available KOS / value vocabularies (LLD.SKOS-LIB-KOS) and links between them (SW.LINKING-INVENTORY).
- Update on the status of library vocabularies in development (LLD.VOCAB-STATUS), including transformation of MARC to RDF, and how to discover these once they’re available.
- Clear statement for the issue of description of Bibliographic entities vs. Real World Entities (LLD.AUTHORITIES, LLD.PERSON-METADATA, LLD.PERSON-NAMES)
- Outreach and dissemination : make sure the XG’s activity is known to any external interested body or organisation, and record initiatives in the wiki (pending action) MGT.OUTREACH
- Demonstrate the need for library linked data, by showing the actual use (USE.END-USERS, USE.COMPU, USE.PRO) - to be extracted from use cases
- Identify new steps, new efforts, future working group, best practices (MGT.NEXT-STEPS, MGT.NEW-EFFORTS, MGT.FUTURE-WGS, MGT.BEST-PRACTICES-GP).
Note : SW.DATA-EXTRACTION could be a tool to present our deliverables.