| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Adam Pease wrote:
> Richard,
> It sounds like you've already looked at WordNet, but it
> has at least
> some of what you may be working on - a list of verbs that
> state actions
> that are implied by the occurrence of an action of the first
> verb. For
> example, killing causes dying
> <http://www.cogsci.princeton.edu/cgi-bin/webwn1.7.1?stage=2&wo
> rd=killing&posnumber=2&searchtypenumber=14&senses=1&showglosses=1>.
> SUMO expands on this a bit since there are axioms that can
> define a whole
> set of things which follow from the occurrence of an action,
> and which are
> more specific than the informal WordNet gloss. A counterpart
> to the above
> WordNet example is in SUMO at
> <http://192.168.200.76:8080/sigma/skb.jsp?req=skb_sx&skb=SUMO_
> skb&term=Killing>
> where the second rule states not only that a Death occurs as
> the result of
> a Killing but that the experiences of the Death is the object
> of the Killing.
>
> Adam
Thanks Adam, it helps to remember that SUMO has added more
content to WordNet, and I'm sure still more will follow.
My focus right now is on toolware that can support the kinds
of ontology results coming out of SUO/SUMO/CYC/IFF. The
underlying structure of the representation is my present
concern while these ontologies are still developing. Hopefully
any tools emerging from this effort can be applied to all
of the above.
Rich
>Richard Cooper wrote:
>
> >Jean-Luc Delatre wrote:
> > >
> > > Richard Cooper wrote in:
> > >
> >
> <http://suo.ieee.org/email/msg09686.html>http://suo.ieee.org/e
> mail/msg09686.html
> >
> > >
> > > [snip]
> > >
> > > Philosophers and metaphysicians are not going to take you
> seriously
> > > if you pretend to tackle the problem only at some syntactic level,
> > > that is, by looking at the common structure of sentences
> > > (yet, above the lexical level of character
> > > codes and XML tags).
> >
> >I intended to define a representation of the words that were
> >used, which is syntactic, along with the several meanings
> >that were used to represent the OAV attributions of the
> >understanding process. So I didn't mean to imply that ONLY
> >the syntactic process is needed. I just wanted to explore
> >a way to represent the sequence of assertions or inferences
> >that are drawn from the utterance, much in the same way
> >that the registry is being discussed as a way to organize
> >the OAV assignments in ontologies and in the IFF.
> >
> >Previously, I've played with this kind of idea in relating
> >dynamically changing OAV assignments in the context of events.
> >Since sentences in a discourse describe a sequence of events,
> >what I'm looking for is a way to organize the OAV history,
> >which is the semantic end product of the understanding process,
> >against the initial sentential forms that lead to the OAV
> >attributions.
> >
> >The reason I find this interesting is that it could provide
> >a sort of IDE for debugging theories of language understanding
> >against available ontologies, when and if they do become
> >available.
> >
> > > PARADOXICALLY that might be the way to go!
> > > Because, (witness the arguments between John Sowa and Stephen
> > > Downes in:
> > >
> >
<http://suo.ieee.org/email/msg09664.html>http://suo.ieee.org/email/msg09664.html)
>
> > reaching an agreement on MEANING, even the meaning of
> > "nothing" is probably hopeless!
>
>Most conversations are dynamic because people are changing
>their word use and syntax to match the understanding process
>of transferring information back and forth among them.
>So although no two of use will ever agree about a large
>dictionary, each one in a conversation maintains a
>dynamically changing dictionary. Conversation changes them.
>
>[snip]
>
> > But, are not we talking to each other at this very moment?
> > And does not those furious postings make SOME sense to everybody?
> >
> > Of course, surely not the same sense to everyone, but there
> > is some SHARED meaning,
> > albeit the "share" could be different and differently
> > distorted for each of us.
>
>[snip]
> >
> > By assuming "similar" meanings of the same word used by
> > different locutors
> > we get perilously close to the "Cyc philosophy" (identical
> > meanings) which is known to fail.
>
>Why do you feel that it is known to fail? Even WordNet
>uses alternative synsets for the same spelling. We all
>use WordNet words in some of those synset forms, as well
>as other forms and words not in WordNet. That doesn't mean
>that WordNet fails. It simply means that the complexity
>of language is so high that we can't find a simple solution.
>This is not unexpected after all the limited success of
>every language understanding project so far.
>
>[snip]
>
> > Cheers.
> >
> > -- Jean-Luc Delatre
>
>Thanks,
>Rich