Recommendation Based on Multimodal Information of User-Item Interactions

In order to alleviate the problem of rating sparsity in recommendation systems, a model called RBMI (Recommendation Based on Multimodal Information of User Item Interactions) is proposed in this paper. Firstly, latent features of items are extracted by convolutional neural network from the description texts of items. Next, for each user, RBMI takes the latent feature vectors and the corresponding ratings of the items that the user have rated previously as the input of a long short-term memory network for learning dynamic latent representation of the user. Finally, the sigmoid function is employed to predict the interaction probability between users and items. Extensive experiments on two real-world datasets have been done and the results show that the proposed model offers better performance and gains significant improvement comparing with the existing state-of- art approaches.