한국어 연상어 사전 구축을 위한 시험적 연구

Abstract

The aim of this paper is to find a methodology for building a Korean associative thesaurus. Associative words are believed to be a vocabulary which frequently co-occurs on a account of their meaning adjacency. Associative words belong to unconscious implicit human memory. Therefore for identifying the substantiality of associative words, it is said to be more efficient to collect them by corpus based approach which can catch frequently co-occurring words rather than to collect them by questionnaire which asks conscious reflection.

On this point of view, this paper attempts to extract associative words from adjacent words to a target word in the corpus. It shows that 75% of associative words collected by questionnaire appear in the adjacent word list, which is made by extracting left and right three words adjacent to a target word. This method also catches associative words extensively, which are not included by questionnaire. It means that corpus based approach to Korean associative thesaurus can be more efficient and more economical. A underlying methodology for building a Korean associative thesaurus is established.