Ranking scientific publications with similarity-preferential mechanism

Along with the advance of internet and fast updating of information, nowadays it is much easier to search and acquire scientific publications. Identifying the quality of papers is thus becoming more and more difficult. Accordingly, many ranking algorithms are proposed to mine high quality articles and quantify scientists’ academic impact. One of the most famous methods is the PageRank algorithm which was originally designed to rank web pages in online systems. In this paper, we introduce a preferential mechanism to the PageRank algorithm when aggregate resource from different nodes to enhance the effect of similar nodes. In other words, a node tends to receive resource from downstream nodes that are similar to it. The method is denoted as Similarity-Preferential Rank (SPR). The validation of the SPR method is performed on the data of American Physical Society (APS) journals. The results indicate that the Similarity-Preferential mechanism improves the performance of the PageRank algorithm in terms of ranking effectiveness, as well as robustness against malicious manipulations. Specifically, SPR can improve the ranking of the high quality papers (e.g. Nobel Prize papers). SPR also significantly outperforms PageRank in predicting the future citation growth of papers, especially for long-term future prediction. Finally, when some papers maliciously cite a target paper to push up its citation, the SPR method can effectively suppress the ranking of this target paper. Though the SPR method is only applied to citation networks in this paper, we believe that it can be naturally used in many other real systems, such as designing search engines in the World Wide Web and revealing the leaderships in social networks.