Linked Open Data: Call For Ideas

An area related to Linked Open Data in which we at UNC-Chapel Hill believe
DLF can most productively contribute is in facilitating actual
implementations that support the solving of problems libraries have. These
will necessarily be experimental at first, but need to move to
production-ready services very quickly in order to be seen in this
community as more than a fad or theoretical movement. A number of library
organizations are publishing LOD, so we believe DLF’s resources are best
spent taking the next step beyond that. We see two general types of
implementations that DLF could support, perhaps in phases – using LOD that
someone else publishes, and building infrastructure to facilitate using
LOD from diverse sources.

For using LOD published by a third party, this would essentially be
bringing home to the “everyday” library what it means to use LOD published
by someone else in their daily work, to accomplish tasks they already need
to accomplish. We see authority data in support of resource description as
an area ripe for this approach. An implementation that employs this
approach might be hacking an ILS (open source or otherwise) to swap out
data from id.loc.gov or VIAF instead of a local authority module. Another
option might be to connect a non-MARC metadata editor such as one of the
XForms-based MODS editors to a LOD data source for fields appropriate for
authority control using that vocabulary. An extension of this approach
might be creating, storing, and publishing LOD locally for authority
records an institution needs to “add” to the vocabulary published by
someone else, and making the two vocabulary sources seem integrated to the
human metadata creator.

For building infrastructure to facilitate using LOD from diverse sources,
this would be the library community imagining and building services that
would translate the infinite and distributed nature of an RDF graph into a
form that could actually be used by production applications. This
infrastructure would need to provide registry services – applications
needing data of a specific type (for data about resources in a certain
class, in a given knowledge domain, using a given ontology, etc) would
talk to this registry to discover what data sources should be brought into
the application. This infrastructure might also provide caching services
of a sort, allowing use of the data to occur faster than the many SPARCQL
queries that would be necessary to discover and retrieve this data from
the source in real time. There would need to be significant exploration of
both the technical and policy (what’s the model for figuring out what data
sources are represented in the registry?) issues surrounding the
development of this infrastructure. We believe place name (geospatial)
data is ripe for testing this approach – making connections between place
name vocabularies from OCLC, LC, the Getty, and GeoNames, and facilitating
their formal expression as geospatial coordinates to promote integration
into mapping services, for example.

For either of these two general types of implementations, some sort of
camp/brainstorming event focusing on that type of implementation
specifically, with representatives from interested institutions, seems
like a reasonable next step. The attendees at the meeting would need to be
diverse enough to ensure that there is substantial knowledge of
initiatives happening beyond the library community related to LOD, but
also “new” enough so that it’s not just the same individuals that are
already active in this area talking to one another. The expectation would
be that that attendees would refine ideas for more specific and targeted
work to follow the planning meeting. DLF could sponsor this meeting along
with specific cooperative initiatives that arise from ideas at the
meeting. Some specific outcomes that would be beneficial from these
follow-on initiatives:

- code libraries
- reference implementations
- partnerships with other players in this space, both organizationally
(NISO, OCLC, LC, etc) and topically (geospatial community, etc)