A New Electronic Dictionary with Meaning Description of Case Frame
Kouhei Shimizu, and Masafumi Hagiwara
Department of Information and Computer Science, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan
In this paper, we propose a new electronic dictionary that helps computers to understand languages. In the field of natural language processing, meaning analysis is a difficult problem. The research has been quite few. On the other hand, meaning comprehension is quite interesting and expected to contribute to many fields. For meaning analysis, knowledge about words is indispensable. Thesauruses provide such knowledge by classifying words by their concepts. They can be used as electronic dictionaries. However, the contents are merely the clusters of words. They have little knowledge about the relationships of multiple clusters, resulting in no capacity for grasping relative meaning of words. For this reason, thesauruses are insufficient for meaning analysis. A novel approach is presented in this paper for improving the capability of electronic dictionaries. The contents of the new electronic dictionary are set of nodes and links, the same style as thesauruses. Our method enables computers to treat concretely multiple concepts of words by deriving nodes from another node. Each node consists of a verb and some nouns. Thus, a node-itself can describe meanings as a form of sentence, which differs from thesauruses. Meaning comprehension is achieved by routing nouns through nodes connected with links. Experimental results indicate the excellent ability to analyze meaning of a whole sentence. It is concluded that such ability of the dictionary is quite useful not only as an electronic dictionary itself but also in many fields such as composition support system.