Dear Jonathon,
The Word Sketch Engine (http://www.sketchengine.co.uk - self-register for a
trial account) has this kind of information available. It also has an API
for accessing the information according to your specifications.
It'll take a bit of thinking about to work out exactly what you want - you
could work simply from the MI of "drunk person" and "drunk spoon", but a
more general solution, less subject to data sparseness, would also use the
built-in thesaurus. There are also possibilities for using a semantic
tagger such as the one developed at Lancaster
http://www.comp.lancs.ac.uk/computing/research/ucrel/usas/
to either enrich corpus data to support more specific queries, or in other
ways,
Regards,
Adam Kilgarriff
-----Original Message-----
From: owner-corpora@lists.uib.no [mailto:owner-corpora@lists.uib.no] On
Behalf Of msc33jmw@cs.bham.ac.uk
Sent: 14 June 2005 10:57
To: CORPORA@UIB.NO
Subject: [Corpora-List] Lexicon with semantic features needed
Dear all
Does anyone know of a freely-downloadable lexicon of words tagged according
to their semantic features? For my purposes (an attempt at auto-generating
cryptic crossword clues) neither the lexicon itself nor the sets of
features need be huge; what I need is a way of evaluating whether two or
more selected words "sit comfortably" together (e.g. DRUNK PERSON) or not
(e.g. *DRUNK SPOON).
Any other potential solutions will be equally welcome!
Many thanks
Jonathan
This archive was generated by hypermail 2b29 : Wed Jun 15 2005 - 07:57:02 MET DST