voula@utcsrgv.UUCP (Voula Vanneli) (02/01/85)
*** REPLACE THIS LINE WITH YOUR MESSAGE ***
ACTIVITIES FOR THE WEEK COMMENCING FEBRUARY 11, 1985
(SF = Sandford Fleming Building, 10 King's College Road)
Dept. of Computer Science
University of Toronto
Artificial Intelligence Seminar - Tuesday, February 12, 3
p.m., SF 1105
Toyoaki Nishida
Department of Information Science, Kyoto University
and
Artificial Intelligence Project, Yale University
"The Application of Montague Semantics to Natural Language Processing"
Abstract: Many natural language processing systems have
been constructed in Artificial Intelligence or Computational
Linguistics. However, it is not always clear exactly what
sort of logical operations are involved in solving linguis-
tic problems. The theory exists implicitly behind a complex
program or grammar. This opacity prevents us from under-
standing, improving, or building a new system on top of pre-
vious systems.
Montague semantics gives a clear view of the process of
interpreting linguistic expressions in terms of abstractly
defined world models. However, it is also true that this
theory should be improved both linguistically and computa-
tionally.
I will present our approach to adapting Montague semantics
to natural language processing by taking computational con-
sideration into account. I will concentrate on two particu-
lar research efforts conducted by our group: generating
semantic networks from English and English-Japanese machine
translation. Based on these experiences, I will talk about
the future development of a more powerful and clean computa-
tional formalism.