A Self-Evaluating Collaborative Filtering Recommendation Algorithm: A Case Study of Anime Recommendations

Abstract

This paper introduces a collaborative filtering recommendation algorithm aimed at addressing the issues of information insufficiency and the mismatch of user personalization needs in anime recommendations. Firstly, we review relevant literature to explore the application of collaborative filtering algorithms in recommendation systems and previous research findings. Then, we detail the design and implementation of collaborative filtering algorithms based on anime and user data, calculating similarities between anime and between users respectively for recommendation purposes. Finally, a self-evaluation is conducted to achieve optimal recommendation performance. Experimental results show that when recommending 10 anime titles, the ROC analysis results indicate a high level of precision, with both algorithms performing well in terms of accuracy.