A Graph-Based Novelty Research on the Music Recommendation
Ruixing Guo, Chuang Zhang, Ming‐Tsang Wu, Yutong Gao · 2016
Nowadays, with the exponentially growth of information, more and more recommendation systems are used for commercial purpose. Among all kinds of recommendation systems, Collaborative Filtering-based Recommender is the most popular one. It is used in a wide range of music recommendation systems and its accuracy is pretty high. However, it is hard for this kind of recommender to find novel things. In this paper, we propose a graph-based novel framework of music recommendation which creates the preference directed graph and the positive correlation undirected graph. With the combination of two graphs, use entropy to get an accurate and novel list. Finally, we compare a traditional collaborative filter algorithm-UBCF with our graph-based novel algorithm-PPGB on the dataset provided by Douban Music. The result shows that PPGB has made great progress.