Room-based Playlist Arrangement System using Group Recommendation

In a room-based environment, to arrange music to meet the preferences of people with uncertain presence, we propose a playlist arrangement system based on group recommendation by singular value decomposition. In group recommendation, we consider uncertain groups with three time-related definitions, historical, current, and predicted presence, enabling the system to provide more diverse playlists. The performance of system is experimented with synthesis data from two real-world datasets. In the pre-processing of data, simulation of time period for location domain dataset and conversion from implicit data to ratings for music domain dataset are used to obtain data needed. The result shows that the proposed system performs well upon providing music with high rating than popular music.