Clustering verb, adjective, adjectival-verb concepts using proximity relation

This paper deals with clustering verb, adjective, epithet-verb concepts, using fuzzy proximity relation, in the view point of noun connection which is called the Theme-Rheme structure. Similarity relation S is denoted between two verbs which own same nouns. That is why similar verbs connect same nouns. After denoting S to all verbs, /spl alpha/-cut the S and make similar groups among the verbs, adjectives and epithet-verbs. The groups and singular verbs consist of verb classes. In a similar way, S is denoted between two adjectives and between two epithet-verbs, and makes adjective classes and epithet-verb classes. 131 to 194 verb classes, 5 to 34 adjective classes and 39 epithet-verb classes are acquired. This research is based on the construction of the situation database and Japanese ideational semantic network in systemic functional grammar.