Item Recommendation Based on Heterogeneous Information Networks with Feedback Information
Yujiao Wen, Fushen Sheng, Ruixue Li, Bangzuo Zhang, Guozhong Feng, Xiaoxin Sun · 2019
In recent years, the heterogeneous information networks have become more and more widely used in recommender systems. There are many conventional recommendation methods. However, most of them only use rating information and the data sparsity problem will reduce the recommendation performance. In this paper, we propose the similarity method combining feedback information with heterogeneous information networks, called SF-H, to predict the rating scores of users on items for recommendation. First, the conventional Pearson Correlation Coefficient is improved by considering the user's additional information. And the Improved Pearson Correlation Coefficient is combined with SVD to obtain feedback information. Then, we combine the feedback information and the heterogeneous information to improve the quality of recommendations. Finally, comprehensive experiments based on two data sets verify that our proposed method performs better than other state-of-the-art approaches in common recommendation tasks.