Sergei: (01)
> For over a decade it has seemed to me that I
> am almost the only one still pursuing ontological-semantic-based NL
> applications (other people did some descriptive semantics but few
> worked on full-blown text analyzers). I think also that -- NL being a
>preeminent
> means of communication with humans -- some form of an NL interface to
> ontological knowledge and its derivatives is essential for practical
> success of the ontological enterprise as a whole. (02)
poor Sergei! :-) Don't forget that my groups use ontologies just
as much now for all our natural language work as we did
15-20 years ago when we were all starting out with
ontology-rich NLP. And I
have never had the impression that natural language interfaces
for Cyc were not an issue! :-) There is also very active work
in many research institutes in Europe where deep processing
is accepted as crucial *and* this is increasingly using
an ontology of some kind or other (and saying loudly that they
do so: cf. all the recent big projects from the DKFI). So
I wouldn't feel too lonely if I were you. (03)
I also certainly agree with you that the evaluations for NLP that
circulate are automatically built so that deep processing is
disadvantaged. And would therefore support any ideas that
you come up with for a more balanced definition of evaluation
tasks. Unfortunately I'll only be remote at the meeting
next week, but am very interested if you manage to come
up with anything. (04)
I'll of course be mentioning the natural language processing
aspects and their potential value in, around and on ontology
efforts in my presentations next week. (05)
Best,
John.
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uos-convene/
To Post: mailto:uos-convene@xxxxxxxxxxxxxxxx
Community Portal: http://ontolog.cim3.net/
Shared Files: http://ontolog.cim3.net/file/work/UpperOntologySummit/uos-convene/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UpperOntologySummit (06)
|