W3C home > Mailing lists > Public > public-swbp-wg@w3.org > August 2004

RE: WNET: first draft of note in WG web space

From: McBride, Brian <brian.mcbride@hp.com>
Date: Thu, 5 Aug 2004 16:49:10 +0100
Message-ID: <E864E95CB35C1C46B72FEA0626A2E8080398513D@0-mail-br1.hpl.hp.com>
To: Guus Schreiber <schreiber@cs.vu.nl>, "McBride, Brian" <brian.mcbride@hp.com>
Cc: public-swbp-wg@w3.org

Hi Guus,

Thanks for the comments on ... 

http://www.w3.org/2001/sw/BestPractices/WNET/wordnet-sw-20040713.html

[...]

> 1.
> 
> [[
> ... words are represented by resources of type wn:Word. The 
> properties 
> wn:hasWordForm and wn:hasLanguage relate a word to its word form and 
> language respectively.
> ]]
> 
> We probably should document why we decide to adopt this 
> instead of using 
> rdfs:label with a language tag. Even in a pure "as-is" approach one 
> might one to use such RDF constructs.

I'm not sure that we've actually got a decision in this as opposed to a
proposal.  However, ...

How much rational to include in a spec is always an issue.  I tend to the
view that one should separate specification and rational because the
rational can too easily weaken the clarity the spec, but that is just a
personal view.  Also the text you quote is in the introduction where I think
the emphasis should be on description rather than justification.

I'd be inclined to put the rational in the more detailed description of the
word class when I get round to adding the property definitions that aren't
there yet.  Shall I do that and see if it addresses your comments?

I was also thinking of having an issues appendix, at least for the
development of the document and keeping track of decisions with rational
there.

> 
> 2.
> 
> [[
> A wn:WordSense resource can be thought of as representing a 
> (word sense, 
> synset) pair.
> ]]
> 
> I assume you mean "*word*-synset pair". wn:WordSense is (in old ER 
> terms) a "weak entity", i.e. its existence depends on other 
> entities. It 
> is linked to precisely one wn:Word and precisely one wn:Synset, right?

Right fixed.

> 
> 
> 3.
> 
> [[
> A word sense, which is sense number N in the synset with 
> identifier $$ 
> is named by $WNBASE/sense/$$/N#
> ]]
> 
> The schema in the document does not talk about a sense number. In the 
> RDF file I see:
> 
> [[
> <rdf:Property rdf:ID="tagCount"> <!--Brian: added -->
>      <rdfs:domain rdf:resource="#WordSense"/>
>      <rdfs:range  rdf:resource="&xsd;nonNegativeInteger"/>
>    </rdf:Property>
> ]]
> 
> Am I correct in assuming this represents the sense number? 
> Can it be a zero?

This was out of data anyway, and even more so now since Danbri pointed out
that WNET now has sense identifiers.  So I think this will change.  Added a
marker for now.

> 
> 
> 4 [silly]
> 
> I guess most nonnative speakers will not know the verb "to dog".

I've switch this to "plant".  Is that better?

I've also added a comments/issues section that lists TF members actions.
Would you rather track these centrally as part of the WG action list?

Brian
Received on Thursday, 5 August 2004 12:35:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:09:39 UTC