Kansei Retrieval Agent Model with Fuzzy Reasoning

Abstract

We propose a Kansei retrieval agent (KRA) model with fuzzy reasoning as the basis for a Kansei retrieval system. In our system, the KRA learns user preferences on the basis of user evaluation of items from a large database. The system employs fuzzy reasoning for the KRA model to express user preferences by using the if–then rules and obtains user preferences using linguistic information. The proposed method optimizes membership function parameters, i.e., the center values and kurtosis of fuzzy reasoning, via user evaluation of various items by using a genetic algorithm. We performed a numerical simulation to demonstrate the effectiveness of the proposed method. In the simulation, we used pseudo users instead of real users and examined the evolutionary performance of the KRA with the proposed method. The results showed that the proposed method was effective in learning user evaluation criteria.