Word-Sense Disambiguation Using Decomposable Models

Word-Sense Disambiguation Using Decomposable Models Rebecca Bruce and Janyce Wiebe Computing Research Lab and Department of Computer Science New Mexico State University Las Cruces, NM 88003 rbruce@cs.nmsu.edu, wiebe@cs.nmsu.edu Abstract Most probabilistic classifiers used for word-sense disam- biguation have either been based on only one contextual feature or have used a model that is simply assumed to characterize the interdependencies among multiple contextual features. In this paper, a different approach to formulating a probabilistic model is presented along with a case study of the performance of models pro- duced in this manner for the disambiguation of the noun interest. We describe a method for formulating proba- bilistic models that use multiple contextual features for word-sense disambiguation, without requiring untested assumptions regarding the form of the model. Using this approach, the joint distribution of all variables is described by only the most systematic variable inter- actions, thereby limiting the number of parameters to be estimated, supporting computational efficiency, and providing an understanding of the data. Introduction This paper presents a method for constructing prob- abilistic classifiers for word-sense disambiguation that offers advantages over previous approaches. Most pre- vious efforts have not attempted to systematically iden- tify the interdependencies among contextual features (such as collocations) that can be used to classify the meaning of an ambiguous word. Many researchers have performed disambiguation on the basis of only a single feature, while others who do consider multiple contex- tual features assume that all contextual features are either conditionally independent given the sense of the word or fully independent. Of course, all contextual fea- tures could be treated as interdependent, but, if there are several features, such a model could have too many parameters to estimate in practice. We present a method for formulating probabilistic models that describe the relationships among all vari- ables in terms of only the most important interdepen- dencies, that is, models of a certain class that are good approximations to the joint distribution of contextual features and word meanings. This class is the set of de- composable models: models that can be expressed as a product of marginal distributions, where each marginal is composed of interdependent variables. The test used to evaluate a model gives preference to those that have the fewest number of interdependencies, thereby select- ing models expressing only the most systematic variable interactions. To summarize the method, one first identifies infor- mative contextual features (where "informative" is a well-defined notion, discussed in Section 2). Then, out of all possible decomposable models characterizing in- terdependency relationships among the selected vari- ables, those that are found to produce good approxima- tions to the data are identified (using the test mentioned above) and one of those models is used to perform dis- ambiguation. Thus, we are able to use multiple contex- tual features without the need for untested assumptions regarding the form of the model. Further, approximat- ing the joint distribution of all variables with a model identifying only the most important systematic interac- tions among variables limits the number …

Content truncated to 500 words