Leveraging semantic features for recommendation: Sentence-level emotion analysis

Abstract

Personalized recommendation systems can help users to filter redundant information from a large amount of data. Previous relevant researches focused on learning user preferences by analyzing texts from comment communities without exploring the detailed sentiment polarity, which encountered the cold-start problem. To address this research gap, we propose a hybrid personalized recommendation model that extracts user preferences by analyzing user review content in different sentiment polarity at the sentence level, based on jointly applying user-item score matrices and dimension reduction methods. A novel voting mechanism is also designed based on positive preferences from the neighbors of the target user to directly generate the recommendation results. The experimental results of testing the proposed model with a real-world data set show that our proposed model can achieve better recommendation effects than the representative recommendation algorithms. In addition, we demonstrated that fine-grained emotion recognition has good adaptability to a sparse rating matrix with a reasonable and good performance.