Applying Keyword Map Based Learner Profile to a Recommender System for Group Learning Support
Xin Wan, Qimanguli Jamaliding, Fumihiko Anma, Toshio Okamoto · 2010
In general, content-based recommender systems use a keyword vector to locate recommendations. However, this method does not consider relations of each keyword and it is also inscrutable to users, who may have a hard time determining which words in their profiles are important and which may be skewing their results to irrelevant recommendations. In contrast, the method proposed in this paper automatically creates a keyword map based user profile for each learner, based on visited learning materials and the learning processes in a web based learning system. The keyword maps describe each learner's existing knowledge with keywords and various relations of them. Our recommender system makes use of this user profile to suggest learning materials the learner might be interested in. In this paper, we report on our initial work on applying keyword map-based learner profile to a content-based recommender system. We believe that our method would provide good accuracy while avoiding many of the problems of both collaborative and keyword based approaches.