基于深度特征融合的协同推荐算法OA北大核心CSTPCD

Deep neural networks faces challenges such as data sparsity and low recommendation accuracy.Therefore,a collaborative recommendation algorithm based on deep feature fusion is proposed to improve the problem by integrating deep neural networks with collaborative filtering algorithms.Firstly,a quadratic polynomial regression model is used to extract features from a user item rating matrix;secondly,using deep neural networks to train the input latent features and generate user item ratings;finally,the recommendation candidate set generated by the term frequency-inverse file frequency algorithm is used to fuse user item ratings and output the recommendation results.Using MovieLens rating data for experiments,the mean absolute error(MAE)and root mean square error(RMSE)of the hybrid recommendation algorithm in this paper are 0.745 9 and 0.888 6,respectively,which are 14.143%and 24.341%higher than traditional deep neural networks,and also better than the mixed recommendation model in the control group.