抄録
When people carry out collaborative works, conceptual difference due to different backgrounds and knowledge can hinder their communication and deteriorate collaboration. We have been carrying out research on detecting conceptual difference by focusing on the situation in which different symbols are used to denote the same meaning and the same symbols are used to denote different meaning. In our approach each user’s knowledge is represented as a decision tree respectively and difference in concept is detected as difference in the structure of decision trees. This paper points out some problems in constructing a unique decision tree based on a single information criterion. Based on the idea of diverse structure, this paper proposes a new method for increasing the performance of detection by constructing multiple decision trees with diverse structure. Genetic algorithm is utilized to realize the idea of diverse structure and experiments with motor diagnosis cases on the implemented system confirmed the improvement on the performance of detection.
