A Computational Method for Resolving Ambiguities in Coordinate Structures

Language, Information and Computation (PACLIC12), 18-20 Feb, 1998, 263-270 A Computational Method for Resolving Ambiguities in Coordinate Structures Haodong Wu and Teiji Furugori Department of Computer Science University of Electro-Communications 1-5-1 Chofugaoka, Chofu, Tokyo 182, JAPAN IwuJurugorij@phaeton.cs.uec.ac.jp Abstract This paper describes a method for determining syntactic structure in coordinate con- structions. It is based on the information taken from semantic similarities, selectional restrictions, and some other linguistic cues. We discuss the role the information plays in resolving ambiguities that appear in coordinate constructions, describe the means of acquiring the necessary information automatically from two on-line corpora and a lexical database, and devise two algorithms for disambiguating coordinate constructions. An ex- periment that follows shows effectiveness of our method and its applicability to resolving ambiguities in some other syntactic structures. 1 Introduction Syntactic ambiguity appears, among others, in coordinate constructions. It is an annoying problem in analyzing structure and meaning of a sentence. A parser, for instance, is to detect the scope of a coordinate structure and identify its inner modification relations. However, the current parsers (e.g., the Link parser) often fail to handle the problem and/or produce a large number of parses. There are a few computational studies that have tried to resolve ambiguities in coordinate constructions (e.g., Paritong, 1992; Cooper, 1991; Bayer, 1996). For example, Kurohashi and Nagao (1994), in analyzing long Japanese sentences, proposed a syntactic analysis method for detecting conjunctive structures by using lexical similarity and structural parallelism. Mela and Fouquere (1996) used a direct process to determine the scope of a coordinate structure based on the concept of functor, argument and subcategorization. Unfortunately, neither of them has sufficiently dealt with .the syntactic structure of a coordination especially when a coordinator (such as and, or and comma) has two or possibly more preceding and succeeding constituents. We in this paper propose a method for determining the structure of a coordinate construction using information on similarities, selectional restrictions, and oilier linguistic cues. In Section 2 we identify the problem and describe the ideas behind our method in Section 3. We give disambiguation algorithms, show an disambiguation experiment, and evaluate its results in Section 4. In Section 5 we suggest an applicability of our method to resolving other syntactic ambiguities. 2 Modification Relation in Coordination Resolving ambiguities in a coordinate construction is to determine the way of conjoining con- stituents (words, phrases, or clauses) and/or to determine the scope of coordination, i.e., imme- diacy relations among the constituents involved. For instance, in sweet and sour pork, the right immediacy relation is ((sweet and sour) pork) rather than (sweet and (sour pork)). Consider other examples. (1) Tom is a ((stock and estate) keeper). (2) John is a (student and (chess player)). (3) Old men and women were left at the village. In each of these sentences, a noun or an adjective that appears in the left hand side of coordi- nator(s) has two (or more) modificant candidates: it may or may not modify the head noun in the right hand …

Content truncated to 500 words