Abstract
We propose Kansei Retrieval Agents (KRA) model with fuzzy reasoning for Kansei retrieval system. Kansei retrieval system searches user's preference items from a huge database. In the system, KRA learns user's preference by user's evaluation of items from a huge database. We use fuzzy reasoning for KRA model to express user's preference by If-then rules and obtain the user's preference with linguistic information. The proposed method optimizes membership functions parameters center value and kurtosis of fuzzy reasoning using user's evaluations of various items by genetic algorithm (GA). We performed a numerical simulation to demonstrate the effectiveness of the proposed method. The simulation verified the number of presented data and the number of data features. The simulation results show that the proposed method is effective for learning user's evaluation criterion.
