Research On Education Resources Recommendation Technology

With the development of Internet technology, the application of education recourses ranges from campus to the Internet. Since MOOC (Massive Open Online Courses) which is on behalf of the online education recourses emerged on the Internet, more attention has been paid to the development of online education resources. Not only can online education resources provide the user more opportunity to learn knowledge, but also bring new challenges, for instance the problem of "information overload". However, most education recourses studies focus on the campus application or traditional online learning website. The education recourses recommendation algorithm based on those studies did not notice the new features of user activities which are unique on the online learning platform and the collective intelligence with the background of the Internet.To make up for the disadvantages of those education recourses recommendation algorithm, this paper analyses the new features of users’ leaning behavior and introduce a new way to estimate the user rating based on both user behavior log and Ebbinghaus forgetting curve. At the same time, this paper makes education resources classification based on the topic model. At last, this paper proposes a user interest model and design reasonable education recourses recommendation algorithm.The primary study in this paper are described as the following parts:1) Analyses the new features of interactive behavior on the online education recourses platform. Considering entropy and Ebbinghaus forgetting curve, design the way of allocating weights of different behavior and time.2) Make classification of education recourses using topic model and the tags users allocated, which also calculate the distribution of education recourses on different topics. Together with the user-item rating, design user interest model.3) Using the main idea of collaborative filtering recommendation algorithm, design new recommendation algorithm based on the classification of education recourses As a result, recommend the items which belong to the topics that user has never learned.4) Based on the user-item rating estimation, design new recommendation algorithm which generates the recommended items which belong to the topics that user has learned.5) Implement the designed process and using the datasets from edX and other online education recourses platform to test and verify the advantages of the recommend algorithm that this paper proposed.