Talk:Draft recommendations page

Karen's proposed outline

• Recommendations
  • Assess
    • Identify sets of data as possible candidates for exposure as LD - identify which parts of the data can be excerpted/extracted as LD (not always whole thing)
    • For each set of data, determine ROI of current practices, and costs and ROI of exposing as LD was: 1.1.1 Identify costs of current practices, and costs and ROI to moving of LLD
  • Plan
    • Investigate migration paths was: 1.1.2 Plan for migration to LLD: technical, managerial, and intellectual
    • Foster a discussion about open data and rights was: 1.1.6 Foster a discussion about open data and rights
  • Facilitate
    • Cultivate an ethos of innovation was: 1.1.3 Cultivate an ethos of innovation
    • Identify Linked Data literacy needed for different staff roles in the library was: 1.1.4 Identify Linked Data literacy needed for different staff roles in the library
    • Include metadata design in library and information science education was: 1.1.5 Include metadata design in library and information science education
  • Design
    • Re-design name and subject authorities for linked data was: 1.3.6.1 Recognize the focus of authority data
    • Develop models of library data for the Semantic Web was: 1.3.6.2 Modeling library data for the Semantic Web
    • Emphasize re-use and linking to integrate library data into the Web was: 1.3.2 Promote the use of vocabularies that are well-understood by Linked Data consumers
  • Prepare
    • Develop best practices and design patterns for LLD was: 1.3.1 Develop best practices and design patterns for LLD
    • Develop policies for RDF vocabulary namespaces was: 1.3.4 Develop policies for RDF vocabulary namespaces
    • Create URIs for library resources was: 1.3.3 Create URIs for library resources in good time
    • Identify tools that support the creation and use of LLD was: 1.2.4 Identify tools that support the creation and use of LLD
  • Collaborate
    • Collaborate with Semantic Web community to create or extend LD standards for library data was: 1.2.3 Identify areas where existing library community standards and Semantic Web standards require extension or development to support LLD
    • Work with the Semantic Web community to standardize the notion of a "named graph" was: 1.2.1 Work with the Semantic Web community standardize the notion of a "named graph"
    • Contribute library experience and knowledge to the effort to create a standard for application profiles was: 1.2.2 Help the Semantic Web community define a general notion of "application profile"
  • Curate
    • Apply library experience in curation and long-term preservation to the Semantic Web
    • Preserve Linked Data vocabularies was: 1.1.7 Libraries should help preserve Linked Data vocabularies [TB]
    • Manage and preserve Linked Data datasets was: 1.1.8 Libraries should help curate Linked Data datasets [from: "Benefits"]

re-wording of "Create uris in good time" [GD]

[based on recommendations from Antoine, who said: Some libraries or library organizations should play a leading role organizing the metadata element set space of the library. The organizational pattern could be quite similar to what happening in the current "non-LD" world with for example LC developing/maintaining/hosting MARC. technology does not really change the leading role played by some actors in the field. One change may be in the way this role is played, though: libraries involved in metadata element set development/maintainance/hosting should probably work in a more cross-domain environment, as modelling for LD is more about networking: universities, W3C, publishers, DCMI and other organizations are partners that come to mind... ]

Assign URIs to library resources

Library data cannot be used in a linked data environment if URIs for library resources [?] are not available for use. Institutions that currently manage and maintain library metadata element sets and standard value vocabularies must provide leadership for LLD by assigning URIs for all of the standards they manage. URI assignment can be facilitated by cooperative efforts and through the use of software and sites that have been developed by the linked data community.

Note that application developers and other users of linked data cannot be expected to delay their activities when resources have not been properly identified. These users may of necessity assign URIs themselves outside of the owning institution's control. When owning institutions are not able to provide URIs in good time they should be willing to delegate the effort to reliable parties in the linked data space.

Gordon: I have added the substance of Antoine's comments to the section in the draft recommendations page Draft_recommendations_page#Create_URIs_for_library_resources_in_good_time_.5BGD.5D

Linked data and authority control [GD] [New issue?]

There is an underlying similarity between traditional authority control applied to metadata records and the concept of linked data applied to metadata statements. The identifier or control number of an authority record for a person, organization, or subject is equivalent to a URI for the same entity. The linking of a bibliographic description to an authority heading or access point is the same principle as applied to linked data. The advantages of authority control in improving efficiency and effectiveness by using the same heading for multiple descriptions, maintaining the heading and associated data as a separate activity, and aggregating data for community use, are also advantages for fully-linked data.

Gordon - this overlaps with 1.3.2 Adoption of Appropriate Vocabularies Outside of Library Standards [PM], 1.2.1.1 Library data is expressed primarily as text strings, not identifiers - already on the draft Issues page, and Encourage interoperability between library data and data from other sources [JS] - below.

Extend and augment the practice of authority control [GD] [Corresponding new recommendation?]

Libraries should exploit their experience of authority control by extending it beyond traditional library concepts of headings and controlled access points for authors and subjects to include a wider range of entities such as persons and organizations, places, and measurements. For linked data this amounts to creating URIs and curating namespaces for controlled terminologies, and using appropriate namespaces from other authoritative communities to avoid unnecessary duplication.

Libraries should recognize that specific bibliographic resources also require authority control to allow them to be better treated as the subjects of other resources, and to establish other relationships resulting from digitization and other forms of derivation, succession, compilation, etc. In a linked data environment, the provision of URIs for resources held in library collections allows other communities to reference those resources and provide a chain of linked data to navigate back to them. This allows libraries to benefit from a wider range of resource discovery services aimed at users from other communities.

Gordon - Again, some overlap with Jodi's section below. Also with existing draft recommendations Promote the use of vocabularies that are well-understood by Linked Data consumers [JS], and Aligning library services with externally-produced information (needs consideration).

Jodi's stuff

Text that might belong elsewhere? Jodi Schneider 14:40, 25 April 2011 (UTC) Tomorrow's problems cannot be solved with today's answers; an outlook to the future is needed to weather rapid change. To meet the pace of technological change, libraries must create a culture which encourages and provides resources for experimentation and research. Short-term activities and proof-of-concept projects provide space where failure is an option, while they open possibilities to discover tomorrow's core activities. Embracing failure -- to see the importance and usefulness to "fail but fail fast" -- opens the way to iteratively discovering successful approaches.

Encourage interoperability between library data and data from other sources [JS]

Encourage interoperability between library data and data from other sources [JS]

Open sharing and reuse

Today library data is hard to use outside the library community, and only limited data (e.g. book reviews, book covers) from external communities is incorporated. Yet in a linked data world, library data should be promiscuous: data should flow both into and out of the library from other communities. Underlying structural and vocabulary similarities will facilitate this exchange: Library data will share features with data from many other sources, including cultural heritage, scientific, governmental and humanities data.

Linking by use of shared vocabularies

Making library data "linkable", both internally and to the outside world, is facilitated by the reuse of well-known vocabularies. LLD--library data expressed in RDF--must use and define relationships to common and existing RDF vocabularies such as Dublin Core, FOAF, BIO, GeoNames. Without these relationships library data remains isolated, much as it is today; we do not want an RDF silo, but rather to join the global metadata cloud.

Integration can be very beneficial. For instance, digital humanities research may focus on library objects. The research data about those objects, expressed in RDF as library linked data, can be directly linked to the library descriptions of these objects.

Linking by identification (owl:sameAs)

Another way to link items is to use owl:sameAs to express their sameness, effectively providing a manual deduplication when items are known, from legacy data, to be equivalent. ** [gd: Note that a specific URI1-sameIndividualAs-URI2 statement only has to be made once; that is, only one triple is required for equivalence between two URIs.]