Research On Learning Materials Personalized Recommendation Based On Combination Recommended Technology

With the popularization and development of Internet, network has occupied more and more important position in the people's lives. Network education different from the traditional teaching methords has became the important way to train people,promote the development of education and research. With rapid development of Internet, the redundantion and expansion of information restrict the development of the network. At the time of information increasing, the personalized information service for the Internet users has emerged. Now, the personalized information has obtained significant business results. The personalized information understand the needs and interest of users and recommend commodities to users which they might be interested, this procession imitate sale man selling things. So, recommended technology that used in the personalized recommendation system applicated in the network education can achieve the same good effort in education. Personalized teaching resources could increase the learning autonomy of students and make students studing high-efficiency. Personalized teaching resources has become an important research content in network education.This thesis has research to personalized recommendation system and recommended technology. Particularly, has detailed analysis of collaborative filtering technology, Includeing user-based collaborative filtering technology and Item-based collaborative filtering technology. The major work done is reflected in the following areas:This thesis has analyzed the shortcomings of collaborative filtering algorithms, and proposed the combined recommendation algorithm of collaborative filtering algorithms which is based on item attributes. This combined algorithms has solved the problem of date sparse in the user-item matrix and item cold start of collaborative. Neighbor of similar attributes item which calculate by the item attributes matrix. The unrating item get it's possible-rating that in the user-item matrix, the possible-rating calculate by the neighbor of similar attribute item. Filling the user-item matrix with the possible-rating that the item unrated. Run the item-based collaborative filtering algorithm on the filled user-item matrix. Through the experiment has proved that the recommended results of the collaborative filtering algorithm based on item attribute matrix better than the algorithm of item-based collaborative filtering. According this combination algorithm has designed model for the personalized learing resources recommendation system. Recommend learning resources to students that using multiple recommended way. At last, the recommended model has used in the personalized learing resources recommendation system and achieve the functions which mentioned in the thesis.