感性マルチメディア検索における擬逆行列を用いた一個人適応法

概要

論文の詳細を見る
This paper addresses to the personal adaptation in retrieving multimedia data based on <I>Kansei</I>. As the multimedia retrieval system we have developed uses SD method and factor analysis, evaluating a lot of media data is required in order to retrieve multimedia data according to the user's favor. This will cause heavy burden to the user as well as the system. For the purpose of avoiding the re-application of factor analysis, the method using the matrix operation is proposed. The proposed method uses the factor burden matrix and the factor score one obtained from the matrix of averaged evaluation values. The factor burden matrix for a user is estimated by using these matrices without factor analysis. This will bring the stability to the system. This paper shows that the better retrieval characteristics can be obtained by using the proposed method.