基于深度特征融合的协同推荐算法

Collaborative recommendation algorithm based on deep feature fusion

深度神经网络存在数据稀疏性难题和推荐精度不高的问题,为此提出一种基于深度特征融合的协同推荐算法,通过将深度神经网络与协同过滤算法相融合来改善问题。首先利用二次多项式回归模型对用户-项目评分矩阵进行特征提取;其次利用深度神经网络对所输入的潜在特征进行训练,生成用户-项目评分;最后利用词频-逆向文件频率算法所生成的推荐候选集,融合用户-项目评分并最终输出推荐结果。利用MovieLens评分数据进行实验,该文混合推荐算法的平均绝对差(MAE)和均方根误差(RMSE)分别为0.7459、0.8886,比传统深度神经网络分别提高14.143%与24.341%,也优于对照组的混合推荐模型。

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 0.7459 and 0.8886,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.