Based Integrated Learning Personalized Recommendation System

With the soaring development in computer network technologies and data base technologies, search and download learning resources has become a trend. When learner confront with huge number and various of network resources , the problem of"information overload"and"resources wandering"have become more and more seriously. to center around users recommend information and service what they are interested in can improve the situation efficiently .But the recommendation system based on collaborative filtering technologies confront deficiency of"cold start"and"data sparsity". This paper on the basic of building users'interest model to draw selective ensemble technologies into users'model classify in order to solve the"cold start"problem .Ensemble learning can significantly improve the generalization ability of learning systems is an important research area in the field of machine learning .cobine package filter with collaborative filtering use the similarity of resource composition to improve similarity computing method fill in rating Matrix in order to solve the problem of"data sparsity"."The research in personal intelligence E-learning system based on selective ensemble technologies"is a science and technology research project which is supported by the provincial education department of Jiangxi. The project is to build an personal intelligence E-learning system by ensemble learning technologies .This system can improve the learning effect of E-learning users and improve the utilization ratio and attraction of E-learning system at the same time .This thesis work is a part of this project . It includes:1,Deep research in ensemble learning and personal recommender system, on the basis of building users'interest model to come up with ideas for a three layer user's interest and two refined model ,and use interest oblivious to optimize the model. Decision Tree Algorithm as individual learner to deal with the collected data and combine with the selective ensemble technologies which a simulate organic evolution arithmetic named genetic algorithm is proposed to select individual learners in order to build a better users'disaggregated model for providing the underlying basis of the improvement in collaborative filtering algorithm .2,Research in collaborative filtering algorithm come up with the disadvantage an improved algorithm is proposed .Combine package filter and collaborative filtering, this algorithm use the comparability calculate of resources to form comparability resources Linked list , according to users'scoring can forecast users'unread comparable resources intelligently .With the forecast scoring the matrix can filled in ,solve the disadvantage of traditional algorithm in data sparsity and dramatically improve the collaborative filtering algorithm's precision.3,Design and realize a personal recommender system which based on selective ensemble technologies. This system combine location users'interest,classify character with intelligent recommend algorithm can analyse users'character recommend studying resource and generate studying alternatives intelligently and so on .Through selective learning to get better classify effort in users'character analysis.