Author(s): |
Jie Yu, School of Computer Engineering and Science, Shanghai University, China, 200072 Kege Xie, School of Computer Engineering and Science, Shanghai University, China, 200072 Haihong Zhao, School of Computer Engineering and Science, Shanghai University, China, 200072 Fangfang Liu, School of Computer Engineering and Science, Shanghai University, China, 200072 |
Abstract: |
With the rapid development of Internet, academic database has become more and more important in research field. While facing large amount of resources, users often get confused and the effectiveness of information seeking will be affected. Academic recommendation is put forward to solve this problem and alleviate users’ heavy information burden. It can automatically recommend academic resources to users according to his/her recent research interests. Collaborative filtering is an indispensable implementation technique in academic recommendation. By recommending resources browsed by the users who have similar interest with active user, the effectiveness of recommendation can be greatly improved. In collaborative filtering, how to discover collaborative user is a key issue. This paper presents a novel method of discovering collaborative users who have similar interest with active user based on concept and relation. User profile is built from user’s browsing academic resources and composed of concept and relation. In addition, taking the feature of academic reading into account, threshold of extracting concept and relation is discussed with experiments. In discovering collaborative user, both concept and relation in user profile are considered in computing similarity degree between active user and candidate user, which ensures the accuracy of recommendation. Experiment results demonstrate the validity and effectiveness
of this method.
|