W3C home > Mailing lists > Public > public-esw-thes@w3.org > June 2007

FW: Small report on SKOS validator

From: Miles, AJ \(Alistair\) <A.J.Miles@rl.ac.uk>
Date: Mon, 11 Jun 2007 12:08:53 +0100
Message-ID: <677CE4DD24B12C4B9FA138534E29FB1D02EC04D7@exchange11.fed.cclrc.ac.uk>
To: <public-esw-thes@w3.org>

Hi Antoine, thanks for the feedback.

> For one vocabulary (5200 concepts) the tests are ok, and results 
> compliant with some investigation on integrity I did with the help of 
> SeRQL queries (for Sesame RDF repository) for info, SKOS Basic 
> Integrity Test Case ran in 11371 ms.
> for info, SKOS Thesaurus Compatibility Test Case ran in 917417 ms.
> For a second vocabulary (35000 concepts), well, I hope I've not put 
> the service on its knees :-| A night has passed now and no news!

Yep, unfortunately the service has some scalability issues ... It's due to the inference process used to uncover the integrity problems, it uses rules to compute the transitive closure of skos:broader then queries for any reflexive statements (x skos:broader x) - I think it's the transitive reasoning that slows things down, I don't fully understand it yet - even tho Dave Reynolds from HP labs explained it to me a while ago :)

> Also, a test that I do in my own strategy and that you don't, it 
> seems:
> testing for "redundant" BTs. In my thesauri, I've got things like X BT 
> Y, Y BT Z and X BT Z explicit. This might not be in general an error 
> (and in my cases it is not) but someone might be interested in knowing 
> that for the thesaurus he is validating. this amount to some form of 
> circularity, when considering semantic relationhip in a non-oriented 
> way.

Yes, this is an interesting case. We should raise an issue about the semantics of skos:broader, skos:narrower and skos:related - the discussion of which will also bear on the issue of OWL interoperability (I still have to read the OWL Semantics properly, it's hard finding the week or more I'll need to understand it :)

Cheers for now,

Received on Monday, 11 June 2007 11:09:01 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:45:40 UTC