ナディヌ・ルラ,相沢輝昭
自然言語理解のための
意味ネットワーク上のユニフィケーションの一般化
Abstract:This paper deals with a generalization of the concept of unification using a
semantic network.
Unification is a pattern matching operation of crucial importance in
Artificial Intelligence. First, a less restrictive semantic network was introduced
using a new relation, called similarity relation, on the set of nodes of the semantic
network. Next, the usual unification based on "exact" matching was generalized
as a similarity unification that can treat a kind of "similarity" matching.
As a possible application of this formalism, a more accurate word-to-word
correspondence in different languages was investigated by developing a Prolog
program for words of both English and French.
This work was done as an internship of the first author at ATR Interpreting
Telephony Research Laboratories.