AW: AW: Lexical representations

Hi,

the point is not to "apply some form of NL" to such labels, but
to have a means to categorize labels to meet certain application
requirements, one example is to be able to tag concepts with word stems,
as this is required if you want to classify documents according to
a set of concepts.

This would also be a requirement if any lexical resource such as WordNet
ought to be represented in OWL, there concepts correspond to synsets which
are referenced by several words (which are synonym to each other).

Thus, I'ld want to leave this flexibility in the language.

Raphael
-----Ursprungliche Nachricht-----
Von: Ian Horrocks [mailto:horrocks@cs.man.ac.uk]
Gesendet: Dienstag, 26. Februar 2002 13:33
An: volz@aifb.uni-karlsruhe.de
Cc: Jeremy Carroll; Webont
Betreff: Re: AW: Lexical representations


I liked the revised text. The use made of labels by Applications such
as the one Raphael describes may try to apply some form of NL
understanding to such labels, but that is beyond the scope of OWL. If
calling such labels "lexical representations" suggests that such a
usage is intended, then that is another reason for changing to
"displayable labels". In any case, being able to determine the
language of the label can only help to improve this kind of analysis
can't it? - the way things are now you might be applying an English
based analysis to French labels.

Ian


On February 25, Raphael Volz writes:
> I'm opposed to changing the text from lexical representations to user
> displayable texts.
> Many of our applications indeed rely heavily on lexical representations.
> E.g. we use word stems to provide references from documents to ontological
> entities in our
> conceptual search application. Word stems are lexical representations for
> ontological
> entities but not intended for user display at all. Some subset of lexical
> representations
> that is labels / documentations are intended for human consumption.
>
> Raphael
>
> -----Ursprungliche Nachricht-----
> Von: www-webont-wg-request@w3.org
> [mailto:www-webont-wg-request@w3.org]Im Auftrag von Jeremy Carroll
> Gesendet: Freitag, 22. Februar 2002 10:54
> An: Peter F. Patel-Schneider
> Cc: www-webont-wg
> Betreff: RE: Lexical representations
>
>
> > Unfortunately, I don't think that this does the trick.  An ontology
> > identifier is, I think, a URI reference.  At least that is what the
second
> > requirement appears to be saying.
> >
>
> Ahh.
>
> I think you are saying that the term "ontology identifier" is being used
for
> two different things. In the second req. as the identifier for an
ontology,
> in this req. as an identifier for some object within the ontology.
>
> I didn't feel very comfortable with your "If ..." since that appeared to
be
> weakening the requirement to an optional one (although I don't think that
> was your intent).
>
> How about hacking "same ontology identifier" to be "same identifier of an
> object within an ontology". It's a bit wordy, but I hope it's good enough.
> i.e. the whole section being:
>
>
> [[[
> User displayable labels
> =======================
> The language must support specifying multiple alternative user displayable
> labels for the same identifier of an object within an ontology .
> This can be used, for example, to view the ontology in different natural
> languages.
> ]]]
>
>

Received on Tuesday, 26 February 2002 08:00:32 UTC