Research On Personalized Recommendation System In E-learning Based On Collaborative Filtering Technology

The application of E-learning provides users with unprecedented abundance of learning materials and great flexible learning ways. The web site can simultaneously accept much more learners to access without the constraints of time and space. However, most web-based E-learning system still focus on web site now, and is stereotyped for each student to see the contents that cannot cater to the students' interests. To design and construct a personalized E-learning environment has attracted much attention of researchers. So this paper presents the recommendation system into the E-learning platform, in order to accomplish the purpose of personalized service.Collaborative Recommendation is an effective method to achieve personalized recommendation system. This paper will introduce collaborative recommendation technology into E-learning personalized recommendation system. The main research contents include:(1) user interest modeling. With the explicit feedback to obtain the static user interest, and implicit feedback approach to get dynamic information of user interest. Then we use vector space model to represent the user interest and learning materials, as well as information supplement including the original information to update the user interest model. (2) put forward an improved collaborative filtering algorithm based on nearest neighbor rating matrix for the sparsity problem, and perform experiments to verify the effectiveness and superiority, our experiment results suggested that this method could effectively avoid the user's evaluation value being amplified when the rating matrix is very sparse, the prediction accuracy is much better than traditional algorithm. (3)design and implement the E-LPIRS personalized recommendation prototype system. From the module's structure design, development platforms, database design and system interface, to give the E-LPIRS system a detailed introduction.The improved methods can effectively alleviate the problem of sparsity and improve the quality of recommendation system. The E-LPIRS personalized recommendation system are universal and significant to other application areas's research and design, with a high reference value.