II-UserCF: Optimizing movie recommendation systems based on UserCF improvement algorithm

With the advent of the e-commerce era and the development of computer technology. Recommendation systems are widely used in people's lives. Traditional user-based collaborative filtering recommendations often produce prediction errors because they do not take into account the popularity of the item. In order to improve this drawback, a new collaborative filtering recommendation algorithm based on a combination of traditional usercf, and item importance is proposed. User relevance is first calculated. Then a new value is obtained based on the different popularity levels of the items. The two are combined in order to obtain a new user similarity. To better predict the similarity between users and to produce more accurate recommendation results. Research shows that adding a value related to the popularity of the product can effectively improve the accuracy of the recommendation.