BTM-CF: A Novel Collaborative Filtering Recommendation Algorithm in the Chinese E-commerce

Abstract

E-commerce platforms often utilize Collaborative Filtering (CF) algorithms, which primarily depend on user rat- ings and overlook the valuable information embedded in review texts. This limitation exacerbates the issue of data sparsity. To address this, we propose a new CF algorithm based on the Biterm Topic Model (BTM), termed BTM-CF. By integrating BTM with traditional CF, the proposed approach categorizes review texts into topics, enabling the extraction of nuanced sentiment preferences. Leveraging user similarity calculations, BTM-CF personalizes product recommendations for individual users. Experimental results using the Jingdong shampoo review dataset demonstrate that the algorithm effectively mitigates data sparsity and enhances recommendation accuracy.