A Weighted PageRank for Scientific Paper Ranking
Abstract— Ranking scientific papers is a challenging but important task. This paper focuses on three issues: (1) how to use publication time information of papers to capture the dynamics of an evolving scientific literature network; (2) how to use metadata information of papers to get better ranking results; (3) how to use topic information to more accurately rank a paper with a specific topic. In response to these problems, we propose a weighted PageRank algorithm which uses citations, publication time, topics, authors, venues and other relevant information collaboratively. We conduct experiments on two public datasets. The results show that our method ranks scientific papers more accurately than existing methods and topic-based PageRank vectors can produce more accurate rankings than a global PageRank vector.
Index Terms— PageRank, Topic sensitive ranking, Heterogeneous networks, Topic-based rank
College of Electronics and Information Engineering, Tongji University, CHINA
Cite: Xiao Liu, "A Weighted PageRank for Scientific Paper Ranking," Proceedings of 2018 the 8th International Workshop on Computer Science and Engineering, pp. 136-140, Bangkok, 28-30 June, 2018.