The Research On A Teaching-material-oriented Hybrid Recommendation Algorithm

The Internet has provided users with a large number of teaching materials. Websites with academic lectures have emerged in an inexhaustible variety, supplying a wealth of learning resources. These brought much convenience to users. However, when they enjoy the convenience of the online teaching materials, they often get lost in the overwhelming amount of information resources. Consequently, users have to put in heaps of time and effort to single out teaching materials tailored to their special interest. In order to make users focus on the contents of these teaching materials, rather than waste a lot of time and energy in seeking valuable materials from the flood of information resources, this thesis conducts in-depth research in teaching material recommendation. The research is mainly focused on the following aspects:1. Firstly, this thesis studies the concept of Recommendation System and Teaching Material Recommendation System, and elucidates the design objectives of the Teaching Material Recommendation System. In addition, it explores the concepts of Teaching Material,Knowledge Points as well as the relationships between different knowledge points, structuring Knowledge Points Tree for a course.2. This thesis designs a personalized recommendation algorithm for text teaching material which is based on both the attributes of the knowledge points and contents. In the first place, it preprocesses the Web log data, analyzing users’ behavior patterns. Based on their behaviors in the browsing process, the users’ degree of interest in the Teaching material is calculated. The Teaching Material Interest Degree is then converted to the Knowledge Point Interest Degree, which further filters the TF-IDF-based recommendation lists, producing text teaching material suggestion lists. Test study indicates that combination of the two algorithms leads to good precision and recall rate.3. This thesis proposes an improved collaborative filtering algorithm- Knowledge Point introduced Item and User-based Collaborative Filtering(KPIU-CF algorithm)- for video teaching material recommendation. The improved algorithm both considers the relationshipbetween knowledge points, and effectively alleviates the problem of data sparsity in the user-item rating matrix to obtain more accurate neighbor users. Experimental results show that this algorithm significantly improves recommendation quality and accuracy of the collaborative filtering recommendation system.4. Based on the different types of teaching materials, this thesis designs and implements a Teaching Material Recommendation System that employs the hybrid recommendation algorithm.