A Weighted Multi-attribute Method for Personalized Recommendation in MOOCs

Abstract

In recent years, with the rapid development of MOOC (massive open online course), more and more people get knowledge from the Internet. The big data analysis for MOOC has become a new research direction and it has been a focus that how to recommend individually personalized videos for MOOC users. To date, the most widely used personalized recommendation technology is collaborative filtering (CF) technology. In this paper, we propose a personalized recommendation algorithm---multi-attribute weight algorithm (MAWA) based on CF. Firstly, MAWA calculates separately the weights of the attributes and attribute values of the resources for the target user. Secondly, the two weights of a video are used to get a recommendation value. Finally, the resources with N highest recommendation values are recommended for the target users. The MAWA in this paper makes up for the shortcomings of traditional CF algorithm and it can be shown from the experiment in this paper that the recall rate of MAWA is 28.3% higher than CF, which means that the recommendation results of MAWA is more accurate than those of CF. The contribution of this paper is to weight the attributes and attribute values respectively, which can reflect the users' preferences in both coarse granularity and fine granularity.