An Improved Dynamic Collaborative Filtering Algorithm Based on LDA

Meng Di-fei, Liu Na, Li Ming-xia, Su Hao-long
2021 IEEE Access  
Currently, existing collaborative filtering (CF) algorithms usually use user behavior data to generate recommendations. The calculation of similarity between users is mainly based on ratings, without considering the explicit attributes of users. This paper proposes an improved dynamic collaborative filtering algorithm named hybrid dynamic collaborative filtering (HDCF), which is based on the topic model. Considering that the user's evaluation of an item will change over time, we add a
more » ... function to the subject model and give its variational inference model. In the collaborative filtering score, we use the topic model to generate a hybrid rating for similarity calculation. The experimental results show that the performance of this algorithm is better than the existing algorithms on the MovieLens dataset, Netflix dataset and la.fm dataset.
doi:10.1109/access.2021.3094519 fatcat:6m3iuvqtbjg6bkfbg34inmlmve