Reproduction Study of Item-CF Algorithm Based on Penalty Factor Correction
Chen Shuangyu · Academic Journal of Computing & Information Science · 2024
With the continuous growth of China's e-commerce industry, the user base for online shopping is expanding, and e-commerce platforms are encountering bottlenecks in their development. The issue of information overload has made it increasingly challenging for consumers to locate the specific products they intend to purchase among the vast array of available options. Given the extensive range of products on e-commerce platforms, traditional collaborative filtering methods struggle with sparse user-item matrices, and less popular items face challenges in receiving recommendations. This paper introduces several algorithms, including collaborative filtering and cosine similarity-based algorithms.To enhance the quality of the recommendation system, this paper incorporates users' feedback coefficients on products into the penalty factor. Additionally, by integrating user portraits obtained through cluster analysis, a collaborative filtering algorithm based on penalty factor correction is proposed.