MOOD ADAPTATION METHOD FOR KANSEI PICTURE RETRIEVAL SYSTEM

Abstract

This paper studies on the adaptation of Kansei multimedia data retrieval system to the user’s mood. In this paper, pictures are chosen as multimedia data because pictures are easier to be evaluated than the other types of multimedia data. Happiness and sadness are adopted as moods because it is easier for the subjects to be in these moods than the other moods. First, the effects of moods on the Kansei retrieval of pictures are experimentally examined. The experimental result shows that there are some effects of moods on the impression of pictures. It also shows that there are two action types in retrieving pictures. One is the action tendency emphasizing the current mood. The other is the one canceling the current mood for subjects to be cool. Next, a method of adapting Kansei retrieval system to the user’s mood is proposed. The proposed method uses the personal adaptation method based on the matrix transformation. The method uses two kinds of information obtained through the experiment. The first is the average scores obtained in the moods. By using these average scores, the impression space can be changed according to the mood. The second one is the impression words used in the retrieval. These are the impression words that the subjects want to use in retrieving pictures in the moods. As the impression words used in the retrieval are decided a priori, the retrieval can start only by specifying the user’s current mood. The prototype system based on the proposed method is constructed. The behavior of this prototype system is demonstrated.