Smart-DJ: Context-Aware Personalization for Music Recommendation on Smartphones
Chengkun Jiang, Yuan He · 2016
Providing personalized content on smartphones is significant in ensuring user experience and making mobile applications profitable. The existing approaches mostly ignore the rich personalized information from user interaction with smartphones. In this paper, we address the issue of recommending personalized music to smartphone users and propose Smart-DJ. Smart-DJ incorporates an evolutionary model called Incremental Regression Tree, which incrementally collects contextual data, music data and user feedback to characterize his/her personal taste of music. An efficient recommending algorithm is designed to make accurate recommendations within bounded latency. We implement Smart-DJ and evaluate its performance through analysis and real-world experiments. The results demonstrate that Smart-DJ outperforms the state-of-arts approaches in terms of recommendation accuracy and overhead.