Research on Digital Curriculum Collaborative Filtering Technology based on Feature Group

With the speedy increase of the amount of data shared on the Internet, it is more important to use human feedback when finding information. Although collaborative filtering is a talented method to salvage useful information, unluckily, there is a difficulty in present collaborative filtering. Since a single system is unlikely to serve every type of object with the best performance, or at least it may be too complex in implementation, it is more practical to use various systems, each of which handles a detailed goal. Personalized recommendation system uses knowledge discovery technology to provide users with accurate and reasonable information to win customers. The recommendation routine based on the characteristics of user groups is a mode that has achieved certain success in both research and practice. However, the complexity of this algorithm increases rapidly with the growth of the number of customers. Therefore, in practical applications, facing tens of thousands of users, the service system has to bear a heavy load of computing, which indications to the decline of recommendation efficiency. This paper proposes a feature based recommendation algorithm to solve the problem of poor scalability faced by user based recommendation algorithm. The use of feature based recommendation algorithm can improve the recommendation efficiency and achieve or exceed the recommendation performance of user based recommendation algorithm.