Hybrid Algorithm Based on Content and Collaborative Filtering in Recommendation System Optimization and Simulation

This paper explores and studies recommendation technologies based on content filtering user collaborative proposes a hybrid algorithm filtering. method not only makes use of the advantages but also can carry out similarity matching for all items, especially when items are evaluated by any user, which be filtered recommended to users, thus avoiding problem early level. At same time, this takes advantage When number users evaluation levels large, rating data matrix prediction will become relatively dense, reduce sparsity make more accurate. In way, system performance greatly improved through integration two. On basis algorithm, was proposed. By combining with item features, feature established replace traditional user-item matrix. K-means clustering performed set recommendations were made. The solve algorithm. new projects, it predict who may interested in projects according project characteristics scoring generate push list, effectively “cold start.” experimental results show that plays significant role solving speed bottleneck problems sparsity, cold start, online ensure better quality.