Print

---

Re Andy's point back here about the "linked data repository"

(I chatted with Andy about this today but thought it was worth
highlighting here)

> It'd be interesting (academically at least), and probably 
> non-trivial, to think about what a linked data repository 
> would look like?  ORE [3] is a helpful step in the right 
> direction in this regard.

I suppose it depends what we mean by "repository" :-)

But for an example of an app which provides a "presentation layer" over
several linked data datasets dealing with data on research publications,
it might be worth having a look at

http://www.rkbexplorer.com/

which is the topic of a thread on the W3C public-lod list here
http://lists.w3.org/Archives/Public/public-lod/2009Jul/0064.html

(I get occasional hiccups with the Java graph viewer thingy in Firefox,
but I think that's just some flakiness at my end)

It uses an ontology called AKT

http://www.aktors.org/ontology/

which models people, projects, organisations, concepts, documents etc.
(I haven't looked at it very hard, but I think it takes a pretty simple
view of documents (document has-URL "... .pdf"; document has-URL "...
.doc" etc))

The individual data sources are listed at

http://www.rkbexplorer.com/data/

and each of them can be viewed separately (though the view is a fairly
"raw" "triples" one)

Some of these datasets e.g.

http://kisti.rkbexplorer.com/

themselves use different ontologies, which I think are mapped to the AKT
ontology when they are exposed in the aggregate view.

At least one of those source datasets 

http://eprints.rkbexplorer.com/

is created by in turn pulling data from a set of eprints.org
repositories, I guess by mapping one of the XML formats exposed by
eprints into the AKT ontology.

So e.g. the paper described by this page

http://authors.library.caltech.edu/7161/

is assigned the URI

http://eprints.rkbexplorer.com/id/caltech/eprints-7161

and is described by this HTML page 

http://eprints.rkbexplorer.com/description/caltech/eprints-7161

and this RDF/XML doc

http://eprints.rkbexplorer.com/data/caltech/eprints-7161 

(which are available by content negotiation on
http://eprints.rkbexplorer.com/id/caltech/eprints-7161 )

And you get a view of that same data from

http://www.rkbexplorer.com/explorer/#display=publication-{http%3A//eprin
ts.rkbexplorer.com/id/caltech/eprints-7161}

Anyway.... I think it's a rather nice example of what can be done - and
even where some of the initial data sources may not themselves be fully
grounded in the linked data principles. I'm not saying it wouldn't be
easier if all the data sources _had_ applied those principles from the
start; I'm sure some tedious mapping might have been avoided, for one
thing! :-)

P.S. My knowledge of the system is just based on what I'm reading from
their Web site and from the mailing list discussion, I hasten to add, so
apologies if I'm misrepresenting any of the detail. Les, or some of the
other Southampton folks, might be able to provide more accurate
information.

Pete 
---
Pete Johnston
Technical Researcher, Eduserv
[log in to unmask] 
+44 (0)1225 474323
http://www.eduserv.org.uk/research/people/petejohnston/
http://efoundations.typepad.com/