Constructing e-learning communities of interest based on learner's rating prediction

Current e-learning applications are limited with respect to learning cooperation and communication among learners, because these applications such as on-line courses offered in China often involve large numbers of geographically dispersed students who have diverse learning preferences and different requirements. Constructing learner communities of interest is critical and necessary to implementing cooperative learning in an e-learning environment. This paper proposes a method for community construction which put the learners with similar interest together to form communities. Here, the similarity between two learners is measured by computing the cosine of the angle between their rating vectors. To address the problems of sparsity in the rating data set, a learner's ratings on the learning objects which he has not rated is predicted by the similarity of objects. Experimental results derived from real learner data have shown that this method can organize learners properly and efficiently.