Research on recommendation system based on integrating semantic similarity
Yongjun Zhang, Hongshuai Wen · 2020
The collaborative filtering recommendation algorithm ignores the semantic relationship between the recommendation item and the user, so this paper proposes an improved collaborative filtering recommendation algorithm that fuses semantic similarity of recommended items. This paper aims to improve the performance of the problem of ignoring the relationship between items and data sparsity in collaborative filtering recommendation. First, TransH algorithm was used to extract the relationship between users and movie entities, Second, the similarity between recommended items is calculated and integrated into the similarity matrix of user ratings, which makes up for the fact that the item based collaborative filtering algorithm does not consider the similarity between items. The experimental results show that this collaborative filtering recommendation system based on integrating semantic similarity has good performance in precision, recall and coverage.