Leo and Rich, (01)
Leo
> the semantics and the syntax are associated at each point in the NL
> derivation (in a compositional fashion), though each uses different
> "logics" that are correlated. (02)
That short description summarizes the fundamental Fregean principle,
which has been very successful for implementing parsers for formal
languages. But NLs are not formal languages. (03)
Leo
> This is of course old-fashioned symbolic computational linguistics,
> not new-fangled corpus-based computational linguistics -- though
> of course both styles should be reconciled. (04)
I agree that any method that has had some successful applications
should be taken into consideration. Most of the successes of the
corpus-based methods come from processing large amounts of actual
data -- instead of Chomsky's naval gazing. (05)
But their semantic representations are weak or almost nonexistent. (06)
Rich
> can anyone shed some more light on why how and what you would
> use a labeled deductive system for? (07)
For YAFMFNLP (Yet Another Formal Method For NLP). Following is a
quotation from the last paragraph of the LDS paper by Finger et al. (08)
FKGK
> The interpretation of natural language expressions qua expressions is,
> as before, strictly syntactic -- a set of metalevel procedures on the
> building of logical configurations. (09)
The formal system that devoted the longest, most sustained effort
to this strategy is probably Ron Kaplan's group at Xerox PARC, which
was spun off as PowerSet, which was bought by Microsoft to enhance
the capabilities of Bing, and which was finally disbanded as one
more formal failure. (010)
There were many very intelligent people working on these systems,
who devoted many person-centuries to their development. But none
of them can understand language as well as a 3-year-old child. (011)
I believe that the reason why is, or should be, obvious: a child
(or an adult learning a second language) begins with pragmatics,
then semantics, and finally syntax. Of the three, syntax is of
the most value for understanding fine details. But the main
points can be understood even with badly mangled syntax. (012)
John (013)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (014)
|