W3C home > Mailing lists > Public > public-swbp-wg@w3.org > December 2005

RE: [WN] Fwd: WordNet Namespace

From: Peter Mika <pmika@cs.vu.nl>
Date: Fri, 16 Dec 2005 15:42:13 +0100
To: "'Mark van Assem'" <mark@cs.vu.nl>, "'Michel Klein'" <mcaklein@cs.vu.nl>
Cc: <public-swbp-wg@w3.org>, <schreiber@cs.vu.nl>, <Benjamin.Nguyen@inria.fr>
Message-ID: <005401c6024e$e7c4d920$fb1e2582@fspc055>

Hi All,

I would confirm my support for the proposed solution and extend the list of
advantages:

-- Two redirects might seem a lot, but one is required by httprange-14 and
the other adds the possibility to commit to particular versions of WordNet.
(The solution is similar to the "latest version" link on W3C documents,
which always points to the latest version, marked by a URL with a date
encoded in it.)
-- There is no best practice in the area; what is emerging can serve as a
guideline for managing/hosting other ontologies.
-- I still don't see a solution with hash URIs that could prevent us from
serving the entire ontology at a request for a particular term.
-- Serving up WordNet dynamically instead of statically eases
maintenance/management. Serving WordNet statically would require a single
file for each URI. (!) Again, the dynamic service can be implemented/hosted
elsewhere so this kind of expertise is not required from Princeton.

And now the advantages list is just as long as the disadvantages list :)

Best,
Peter


> Hi Jeremy,
> 
> > In light of the httpRange-14 resolution, the slash solution is not
> > totally trivial, since there is meant to be a 303 response in there
> > somewhere ....
> 
> This means we need Princeton to do a redirect, right? This may be
> possible but we have to check with them.
> 
> > So there are two, related but separable problems:
> >
> > a) what are convenient sets of files for different sorts of users, to
> > use as wordnet in RDF & OWL?
> > b) what does a GET return on each of the wordnet URIs?
> 
> I would like to continue working on (a) with hash URIs as it is now,
> while we discuss solutions on (b). So the next RDF version and draft
> Note might not reflect these discussions yet, although of course I
> will try to be as up-to-date as possible. Also, if it is not possible
> to let Princeton to do redirects it seems sensible to stick with the
> files-with-hash-URIs solution for both (a) and (b).
> 
> I discussed with Peter Mika and Michel Klein about a solution to (b):
> 
> - we use slash URIs in the Princeton space, e.g.
>    http://wordnet.princeton.edu/rdf/entity
> 
> - Princeton server redirects this with 303 to another namespace which
> has the version number of the latest version, e.g.
> 
> - http://wordnet.princeton.edu/rdf20/entity
> 
> - Princeton redirects this with 303 to another namespace where we host
> the actual RDF files. THis can e.g. be a server at my university (VU
> Amsterdam)
> 
> - there a dynamic script queries the current RDF files for all
> subjects that have the requested URI in a triple which are then returned.
> 
> Advantages:
> - compliant with httpRange-14
> - GETting one URI does not return whole WN-RDF file
> 
> 
> Problems/Issues:
> - requires Princeton to do redirect
> - three redirects for one lookup
> - requires dynamic script and hosting other than Princeton
> - we should agree on which triples should be returned (proposal is
> only triples with that URI as subject, other possibilities exist)
> - this is not best practice we're implementing
> 
> Mark.
> 
> --
>   Mark F.J. van Assem - Vrije Universiteit Amsterdam
>         markREMOVE@cs.vu.nl - http://www.cs.vu.nl/~mark
Received on Friday, 16 December 2005 14:45:39 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:15 UTC