On Mar 11, 2006, at 6:29 PM, Bill Andersen wrote: (01)
> Sergei's suggestion is an excellent one.
> Perhaps one very useful outcome of the UO summit would be begin an
> effort to find funding for such tests. Given that application
> development costs are as high as they are, we almost never get the
> kind of comparative studies we're talking about – no business has
> the resources to do them and purely empirical work such as this is
> IMHO somewhat unpopular in the academic world.
> I'm thinking something like the HALO project, except extended in
> the way Sergei suggests (maybe OMG or some related body would be
> interested in backing the "non-ontology" team)... (02)
It would be a pleasure to work on specifying such an experiment.
Most, if not all, of the NLP-related "bake-off" tasks so far have
been designed in such a way that the shallow, semantics-free and
knowledge-lite approaches have a built in advantage.
Different formulations and metrics are possible. Maybe we can have a
"birds-of-a-feather" meeting at the meeting next week to toss around
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 (04)