There are two main flaws in tag-based recommendation algorithms,the lack of quantification of user preference for tags,and the weight of different tags in user usage.To solve these problems,a personalized recommendation algorithm from the perspective of tags was proposed.The tags used in the user's historical behavior were analyzed,the user's tag interest model was established according to the user's historical behavior,and the tag interest model was used to calculate the user's preference for different tags.The user's historical rating records were counted and the share of different tags was calculated.The two were linearly combined to get the user's interest in the label.The cosine similarity was used to calculate the user preference similarity,and the user preference similarity was introduced into the matrix decomposition model for item rating prediction and recommenda-tion.Experimental results show that on the MovieLens dataset,compared with the traditional algorithms LFM and SVD++,the algorithm reduces the RMSE by 5.00％and 1.41％,respectively,and reduces the MAE by 5.07％and 1.00％,respectively.
维普
万方数据