Preference Similarity Analysis of User Preference Rules Using a Character Coordination System

Abstract

In this study, we analyzed user preference rules obtained from a character coordination system using the Kansei retrieval agent (KaRA) model with fuzzy inference for the acquisition of user rules related to user preferences. The preference rules of the character coordination system were expressed by if-then rules. Previous studies have demonstrated that the character coordination system was effective in terms of acquiring preference rules that were the evaluation criteria of users. In this study, we measured the similarity of preference rules by calculating the distance between preference rules. As a result, the proposed system generated fuzzy rules that matched with more than 60% of the users’ preferences. The similarity with antecedent labels (if part) of each was high when the consequent labels (then part) were the same.