Cross-domain recommendation algorithm based on deep fusion of content and implicit feedback
Most of the existing cross-domain recommendation methods use only the rating information and some side information from the source domain,and the other side information including implicit feedback information can not be adopted.Therefore,a cross-domain recommendation algorithm which integrate multiple side informa-tion including implicit feedback information and content information is proposed to improve the performance of cross-domain recommendation methods.Based on the expansion of stacked denoising autoencoder(SDAE),combing with matrix factorization(MF)method and fusing the rating information of the source domain,the con-tent information of users and projects and implicit feedback information are also integrated in the method.On this basis,the cross-domain collaborative filtering framework suitable for the comprehensive application of rating in-formation and multi type side information is designed.In order to effectively transfer the source domain informa-tion,both the codebook-based knowledge transfer method and the incomplete orthogonal nonnegative matrix tri-factorization method are adopted in this framework.The experimental results on the actual data set show that this method has a good effect in improving the recommendation performance and reducing users'aversion to the rec-ommendation results.
side informationimplicit feedbackmatrix factorizationcross-domain recommendation
万方数据
维普
