A Probabilistic Matrix Factorization Recommendation Method Based on Deep Learning

Abstract

In order to improve the accuracy of recommendation, a probabilistic matrix factorization recommendation method based on deep learning(PMFDL) is proposed. The method considers the influence of context information on the implicit feature of items and the influence of time factor on the implicit feature of users. In this paper, a convolutional neural network with attention mechanism is introduced to learn the implicit feature of items, and a long-term and short-term memory network is introduced to learn the implicit feature of users. Finally, we combine probabilistic matrix factorization(PMF) to predict recommendation results. After experimental verification, the experimental results show that the proposed PMFDL method is superior to Probabilistic Matrix Factorization(PMF) and Convolutional Matrix Factorization(ConvMF) in recommendation accuracy.