Music recommendation aimed at improving recognition ability using collaborative filtering and impression words

Music therapy to improve recognition ability may be more effective when the favorite music of each person is adopted. A method is proposed that combines collaborative filtering and music recommendation based on impression words. In the proposed method, once the recommendation process using collaborative filtering is terminated because the number of users is zero in the reference user list of users with the same preference of recommended music, the second recommendation process finds the most similar music from the scores of impression words, so that music is successfully recommended from music not previously recommended. In experiments, 12 users rated 52 songs in a textbook database of songs for elementary schools. The number of recommended songs by the proposed method was 10 per user, whereas that of collaborative filtering was 6.25 per user. The recommendation accuracy of the proposed method was 68.3%, whereas that of collaborative filtering was 66.9%.