Optimization of Weighted Hybrid Algorithm based on Collaborative Filtering and Content Filtering in Steam Game Recommendation

As the number of games on the Steam platform continues to grow, it becomes more and more difficult for users to quickly find the games that match their interests among the huge number of games. This study aims to optimize a set of weighted hybrid recommendation algorithms integrating collaborative filtering and content filtering to improve the accuracy and experience of personalized game recommendations. By writing a crawler program in Python, we collect game data on the Steam platform in real time, and construct an original dataset containing game types, ratings, vendors, and other dimensions. Subsequently, cluster analysis is used to model users' historical purchasing behavior, combined with content filtering algorithm to extract game feature vectors, and collaborative filtering algorithm is applied to calculate user similarity, and finally weighted hybrid strategy is used to generate recommendation results. The research results show that the method has good practical application value in alleviating the cold-start problem and improving recommendation diversity and accuracy.