W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > December 2003

Re: too many positive entailment test results

From: Sandro Hawke <sandro@w3.org>
Date: Mon, 01 Dec 2003 16:26:35 -0500
Message-Id: <200312012126.hB1LQZOH007970@roke.hawke.org>
To: "Jos De_Roo" <jos.deroo@agfa.com>
Cc: "w3c-rdfcore-wg" <w3c-rdfcore-wg@w3.org>

> Sandro - it seems as if the results of the negative entailment
> tests are included in the results of the positive entailment tests
> at http://www.w3.org/2003/11/results/rdf-core-tests
> (I was triggered by my low percentage :))

Ouch.  This was caused by someone's feed claiming each of those NETs
was a PET.  It's fixed now (for now).

I'm not quite sure what lesson to learn from this, or how to avoid it
in the future.

I think all RDF applications (at least with aggregation and/or
inference) need "Justify This Data" links all over (what DanC calls an
"Oh Yeah?!?" button) -- maybe if you'd been able to click on that
link, where it said some of those NETs were PETs, you would have known
who to blame.  Better than nothing, but not perfect.

Alternatively, the test ontology could say NET and PET are disjoint
classes, and my app could check for consistency of data.

Or my app could query each source only for the data it wants from that
source -- so the Agfa feed could only give me test results about Euler
-- but that seems too restrictive.

    -- sandro
 
Received on Monday, 1 December 2003 16:26:33 EST

This archive was generated by hypermail pre-2.1.9 : Monday, 1 December 2003 16:26:36 EST