Rating Refinement and Optimized Clustering for Rating Prediction using Collaborative Filtering

Collaborative filtering (CF) is a typical and widely used recommendation method. The main idea is to recommend to the target user items that users similar to him/her like. CF has two main challenges. One is that users' ratings of items are sparse and the other is how to select appropriate similar users to calculate the recommendation results for target users. To address the former challenge, we use the mean ratings and median of standard ratings to fill in the missing values in the rating matrix. To address the latter challenge, we use the butterfly optimization algorithm to determine the initial centers of K-means clustering, and combine both ratings and genres of items to define the hybrid similarity of users. To evaluate the performance of the proposed method, we selected five competitive algorithms and conducted experiments on three publicly available datasets. The results showed that the proposed algorithm improved prediction accuracy.