Improved Collaborative Filtering for Personalized Movie Recommendations Using K-Means Clustering

This study focuses on personalized movie recommendations, aiming to explore advanced recommendation system algorithms. Utilizing collaborative filtering for initial data processing, the research further employs k-means clustering and combines advanced activation functions, attention mechanisms, multi-head attention, and regularization techniques to enhance accuracy. Building upon these methods, an intelligent recommendation system based on collaborative filtering is developed and optimized. The model achieves an accuracy of 99% when correcting the predicted rating for User 6 on Movie 2. The code is available at https://github.com/yuzengyi/CFPMR.