Mikhail Belkin and I are working along lines that might interest you --
we're primarily interested in learning from very raw corpora. We have a
paper on building a two-dimensional representation -- a map, so to speak
-- for words in a corpus, based on the distribution of words adjacent to
them.
We have a preliminary report at
http://humanities.uchicago.edu/faculty/goldsmith/eigenvectors/
John A. Goldsmith
Department of Linguistics, The University of Chicago
ja-goldsmith@uchicago.edu
http://humanities.uchicago.edu/faculty/goldsmith
-----Original Message-----
From: owner-corpora@lists.uib.no [mailto:owner-corpora@lists.uib.no] On
Behalf Of Yuval Feinstein
Sent: Wednesday, January 23, 2002 4:11 AM
To: Corpora List
Subject: Corpora: Syntactic/Phonologic network?
I know of several attempts to build sematic networks according
to semantic relations between words (e.g. cyc, WordNet). Are there
attempts to do the same for syntactic or phonological information? (e.g.
"send" will be close to "convince" syntactically because both are
ditransitive verbs.
"fish" and "wish" are similar phonologically)
Thanks,
------------------------------
Yuval Feinstein
email : yuvalf-w@cs.huji.ac.il
------------------------------
This archive was generated by hypermail 2b29 : Tue Jan 29 2002 - 02:51:26 MET