The first Real-World Application of Kansei retrieval agent is Character Coordination System. This system searches for a coordination that user might prefer from character database. The database storages many various character coordination. This agent model is consisted by fuzzy inference and has a possibility that can find of user’s potential preference rule by Fuzzy rules. Also, character is made up of eye color, hair color, dress color and pattern, and shoe color.


By basis experiments, we confirmed that the system can extract a certain number of rules that fit to users’ Kansei evaluation.


Further, we will improve the proposed system what be able to more accurate extract.