A Clustering-Based Collaborative Filtering Recommendation Algorithm via Deep Learning User Side Information

Collaborative filtering (CF) is a widely used recommendation approach that relies on user-item ratings. However, the natural sparsity of user-item ratings can be problematic in many domains, limiting the ability to produce accurate and effective recommendations. Moreover, in some CF approaches only rating information is used to represent users and items, which can lead to a lack of recommendation explained. In this paper, we present a novel deep CF-based recommendation model, which co-learns users’ abundant attributes. To better understanding the user’s preference, we explore user deeper and unseen factors on the user-item ratings and user’s side information by adopting the AutoEncode network. After that, we conduct the k-mean algorithm with extracted deep user factors to classify users. Then the user-side CF algorithm is employed to produce the recommendation list based on the classification results, for alleviating recommendation speed. Finally, we conduct lots of experiments on real-world datasets. Compared with state-of-the-art methods, the results show that the proposed method has a significant improvement in recommendation performance, in terms of recommendation accuracy and diversity. Furthermore, it also enjoys high effectiveness, and the approach is useful when it comes to assigning intuitive meanings to improve the explainability of recommender systems.