Idea-deriving Information Retrieval System

This paper presents the information retrieval system that integrates concept-based retrieval, which focuses on the similarity of the meanings of words, with the characterstring-matching-based retrieval. The system provides a word association function, a concept retrieval function, and a document classification function, which are expected to help the user to reach the target document quickly or to derive new ideas. This paper describes this system and evaluates the characteristics of various methods used in the system. The evaluation results show that integrating the concept base and character-string matching gives better results than either of them does singly.