(College of Information, Liaoning University, Shenyang 110036, China)
(Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University, Changchun 130012, China)
[Objective]Improved matrix factorization algorithm, fusion of neighbor comments to alleviate sparse comments and improve recommendation accuracy.
[Methods]The Multi-layer Perceptron is used to improve the matrix decomposition algorithm to obtain the deep nonlinear features of users and commodities; process the review set and deeply integrate the neighbor features and user features to characterize the features that are more in line with user preferences; and make recommendations based on the obtained feature prediction score.
[Results]Compared with other deep models on the Amazon data set, it is found that the accuracy rate, recall rate, and normalized cumulative loss gain can be increased by up to 8.3%, 22.8%, 14.9%.
[Limitations]The time factor of the user's comment is not considered, and the existence of false comments is ignored.
[Conclusions]Improved matrix factorization algorithm and fusion of neighbor comments can effectively improve the accuracy of recommendations.