Learning the Space of Word Meanings for Information Retrieval Systems

Learning the Space of Word Meanings for Information Retrieval Systems Koichi HORI, Seinosuke TODA and Hisashi YASUNAGA National Institute of Japanese Literature 1-16-10 Yutakacho Shinagawaku Tokyo 142 Japan Abstract: Several methods to represent meanings of words have been proposed. However, they are not useful for information retrieval systems, because they cannot deal with the entities which cannot be universally represented by symbols. In this paper, we propose a notion of semanUe space. Semantic space is an Euelidean space where words and entities are put. A word is one point in the space. The meanings of the word are represented as the spaee configuration around the word. The entities that cannot be represented by symbols can be identified in the space by the location tile entity should be settled in. We also give a learning mechanism for the space. We prove the effectiveness of the proposed method by an experi- ment on information retrieval for the study of Japanese literature. 1. Introduction There have been no theories of semantics we can rely on for building a large information retrieval sys- tem. The defect in the existent theories is the lack of explanation of the mechanism for adjusting to tile real world the formal symbolic systems used in the theories; the only tMng they explain is the relation between natural language and the formal system. Those theories assume the existence of fixed and universal one-to-one relations between the basic ele- ments in the formal system and tile entities m tile real world. For example, both Montague semantics and the situation semantics assume that we can represent the dog named Morris in the real world as some symbol like MORRIS in the formal system and that the relation between Morris and MORRIS is fixed and universal [3,2]. However, when we consider an information retrieval system, especially in the field of study on literature, we encounter problems where the assump- tion does not hold. One problem is that there are entities that do not have universal symbolic representation. For example, when a researcher dis- covers a new entity(or notion) in literature and writes a paper on that entity, the paper must be stored in the database but we do not have appropri- ate key words for that entity. When tile entity becomes well known in later years, it may be named, for example, 'overthereism'. However at the time tile entity is discovered and does not have the name 'overthereism', we must represent tile entity by a fixed set of symbols, but it is not easy. Another problem is that the range of what is meant by a sym- bol differs among the users of an information retrieval system. For example, we cannot identify the fixed meaning of 'romanticism'. Every user assumes different meanings of 'romanticism' and it is not easy to control the meaning. The latter problem has been considered in the studies of fuzzy meanings, but, so far, the former problem has not been con- sidered m the studies on semantics. In order …

Content truncated to 500 words