Content-based (CB) and collaborative filtering (CF) recommendation algorithms are widely used in modern e-commerce\nrecommender systems (RSs) to improve user experience of personalized services. Item content features and user-item rating data\nare primarily used to train the recommendation model. However, sparse data would lead such systems unreliable. To solve the data\nsparsity problem, we consider that more latent information would be imported to catch usersâ?? potential preferences. Therefore,\nhybrid features which include all kinds of item features are used to excavate usersâ?? interests. In particular, we find that the image\nvisual features can catch more potential preferences of users. In this paper, we leverage the combination of user-item rating data\nand item hybrid features to propose a novel CB recommendation model, which is suitable for rating-based recommender\nscenarios. The experimental results show that the proposed model has better recommendation performance in sparse data\nscenarios than conventional approaches. Besides, training offline and recommendation online make the model has higher efficiency\non large datasets.
Loading....