A Corpus-Based Approach for Building Semantic Lexicons

A Corpus-Based Approach for Building Semantic Lexicons Ellen Riloff and Jessica Shepherd Department of Computer Science University of Utah Salt Lake City, UT 84112 riloff©cs, utah. edu Abstract Semantic knowledge can be a great asset to natural language processing systems, but it is usually hand-coded for each applica- tion. Although some semantic information is available in general-purpose knowledge bases such as WordNet and Cyc, many ap- plications require domain-specific lexicons that represent words and categories for a particular topic. In this paper, we present a corpus-based method that can be used to build semantic lexicons for specific cat- egories. The input to the system is a small set of seed words for a category and a rep- resentative text corpus. The output is a ranked list of words that are associated with the category. A user then reviews the top-ranked words and decides which ones should be entered in the semantic lexicon. In experiments with five categories, users typically found about 60 words per cate- gory in 10-15 minutes to build a core se- mantic lexicon. 1 Introduction Semantic information can be helpful in almost all aspects of natural language understanding, includ- ing word sense disambiguation, selectional restric- tions, attachment decisions, and discourse process- ing. Semantic knowledge can add a great deal of power and accuracy to natural language processing systems. But semantic information is difficult to ob- tain. In most cases, semantic knowledge is encoded manually for each application. There have been a few large-scale efforts to cre- ate broad semantic knowledge bases, such as Word- Net (Miller, 1990) and Cyc (Lenat, Prakash, and Shepherd, 1986). While these efforts may be use- ful for some applications, we believe that they will never fully satisfy the need for semantic knowledge. Many domains are characterized by their own sub- language containing terms and jargon specific to the field. Representing all sublanguages in a single knowledge base would be nearly impossible. Fur- thermore, domain-specific semantic lexicons are use- ful for minimizing ambiguity problems. Within the context of a restricted domain, many polysemous words have a strong preference for one word sense, so knowing the most probable word sense in a do- main can strongly constrain the ambiguity. We have been experimenting with a corpus- based method for building semantic lexicons semi- automatically. Our system uses a text corpus and a small set of seed words for a category to identify other words that also belong to the category. The algorithm uses simple statistics and a bootstrapping mechanism to generate a ranked list of potential cat- egory words. A human then reviews the top words and selects the best ones for the dictionary. Our ap- proach is geared toward fast semantic lexicon con- struction: given a handful of seed words for a cate- gory and a representative text corpus, one can build a semantic lexicon for a category in just a few min- utes. In the first section, we describe the statistical bootstrapping algorithm for identifying candidate category words and ranking them. Next, …

Content truncated to 500 words