On combining text-based and link-based similarity measures for scientific papers

Abstract

In computing the similarity of scientific papers, text-based and link-based similarity measures look at only a single side of the content or citations. In this paper, we propose a new approach to compute the similarity of scientific papers accurately by combining the text-based and link-based similarity measures. Our proposed method considers the content and citations of the scientific papers simultaneously and combines the similarity scores based on the content and citations by using SVMrank. The effectiveness of our proposed method is demonstrated via extensive experiments on a real-world dataset of scientific papers. The results show that more than 20% improvement in accuracy is obtained with our approach compared with previous methods.