Research on Collaborative Filtering Recommendation Algorithm Based on Sentiment Analysis and Topic Model
Ping Sun, JinShan Li, Guohui Li · 2019
At present, the research about the personalized recommendation algorithm is very popular, there are two problems in this kind of algorithm. First, in the pre-filling stage, user preferences cannot be expressed only by scores, because only using the score-filling matrix has a certain deviation. The second is that in the forecasting stage, the prediction method only relies on the locally relevant scores, which cannot grasp the overall user score features. In order to resolve these two problems to improve the efficiency and accuracy of recommendation algorithm, this paper proposes a collaborative filtering recommendation algorithm based on sentiment analysis and topic model. Experiments show that the proposed algorithm outperforms User-based CF, LDA-CF and BiasSVD algorithms in prediction accuracy.