Music recommendation hybrid system for improving recognition ability using collaborative filtering and impression words

Music therapy for improving recognition ability may be more effective when the favorite music of each person is adopted. In the proposed system, first, the recommendation process using collaborative filtering is terminated when no users in the reference list have the same preference of recommended music as that of a new user. Then, the second recommendation process finds the most similar music, from the scores for impression words, to those successfully recommended among music not recommended up to the moment. The average number of recommended songs for each user by the proposed system was 12.1, whereas that of collaborative filtering was 4.3. The recommendation accuracy of the proposed system was 70.2 %, whereas that of collaborative filtering was 62.1 %. The ratings of songs can be added on a user-by-user basis in the recommendation process, and this increased number of cases improves the recommendation accuracy and increases the number of recommended songs.

We would like to thank Associate Professor M. Tabuse of Kyoto Prefectural University for his valuable advice and support. We would also like to thank all the participants who cooperated with us in the experiments. This work was partially supported by SCOPE (122307003) of Ministry of Internal Affairs of Communications of Japan government.

Authors and Affiliations

1. Works Applications Co., Ltd., 1-12-32 Akasaka, Minato-ku, Tokyo, Japan

   Saya Yoshizaki

2. Graduate School of Life and Environmental Sciences Kyoto Prefectural University, 1-5 Nakaragi-cho, Shimogamo, Sakyo-ku, Kyoto, 606-8522, Japan

   Yasunari Yoshitomi & Taro Asada

3. ISI Software Corp., 4-6-17 Honmachi, Chuo-ku, Osaka, 541-0053, Japan

   Chikoto Koro

Corresponding author

Correspondence to Yasunari Yoshitomi.