- From: Hugh Glaser <hg@ecs.soton.ac.uk>
- Date: Mon, 9 Feb 2009 15:14:12 +0000
- To: Andreas Langegger <al@jku.at>, Juan Sequeda <juanfederico@gmail.com>
- CC: Georgi Kobilarov <georgi.kobilarov@gmx.de>, Yves Raimond <yves.raimond@gmail.com>, "public-lod@w3.org" <public-lod@w3.org>
OK. Let me be a bit optimistic here :-) And plug my technology :-) Despite my frustrations, I did manage to do some linking. I had a quick go at the languages, and the fruits of those labours can be found in the CRS of the courseware.rkbexplorer.com KB. This means that for example users of lingvoj now have an easy way of linking to some other URIs, for example by http://courseware.rkbexplorer.com/crs/export/?term=http://www.lingvoj.org/la ng/bg without the lingvoj people doing the work themselves. (I know that others of the URIs in the equivalence class have already done some linking, and put it in their KBs.) I also did a little bit of place names just to see how it would go, such as http://os.rkbexplorer.com/crs/export/?term=http://sws.geonames.org/2647554/ (We will improve access to the CRS services soon, in case you find the interface indigestible.) I therefore think that if we could persuade some (enthusiastic PhD student?) people to generate the knowledge, we can have the engineering in place to use it. So even if publishers do want to leave it to others, KBs specialised to coreference such as these CRSes are a good engineering solution. And I think other people (okkam, Sindice) are also looking to fill this role. So there is good expectation (more than just hope). Best Hugh By the way, I did download the opencyc equivalence data (as in the message below), with an intention of bringing it up as a CRS, but ran out of time. On 09/02/2009 14:42, "Andreas Langegger" <al@jku.at> wrote: > Georgi, > > d2r-server and Virtuoso RDF views (and others) is only the base technology > which "enables" people to expose RDF out of RDBMS. > However, to use them accordingly I'm afraid Juan and Hugh aren't so wrong. > You'll have to invest a lot of time to use the right vocabularies (a) and > right (external) URIs (b) to expose the right things (c). > > Even internal linking is not trivial. Usually a DB schema is not that what you > want outside. So you have to invest some time for the mapping task. And > linking only internally doesn't contribute much, so adding external links is a > must and that's some effort. > > I don't want to be that pessimistic, but if I've had THE idea to solve this, I > would share it.... > > Regards > Andy > On Sat, Feb 7, 2009 at 8:15 AM, David Baxter <retxabd@gmail.com> wrote: > Hi Hugh, > > The OpenCyc ontology has English strings for its terms, and a search facility > at > > http://sw.opencyc.org/ > > Sorry, no Tim Berners-Lee (yet), but if you type "Tim " it will auto-complete > and show you Tim Duncan and the other Tims we have. > > We also have links to DBpedia for many of our terms. > I neglected to mention that the source of the OpenCyc-to-DBpedia links is a > research project by Olena Medelyan and Catherine Legg (a former > colleague of mine at Cycorp), at University of Waikato, NZ. > The work is described at http://www.cs.waikato.ac.nz/~olena/cyc.html > <http://www.cs.waikato.ac.nz/%7Eolena/cyc.html> . > David Baxter
Received on Monday, 9 February 2009 15:15:22 UTC