Music recommendation system using lyric network

This paper proposes the method to recommend music using lyric network. This method corresponding to more than thousands of musics. The authors focus each lyric of the music. Keywords representing music are extracted from its lyric by combining TF-IDF method and principle of discriminant analysis. Lyric network is generated based on extracted keywords. The connection of generated network can recommend other musics. Numerical experiment is carried out in order to analyze the lyric network constructed by the proposed method and investigate the effect on music recommendation. Experimental result shows the extraction for collection of musics whose lyrics are similar.