Focused Page Rank in Scientific Papers Ranking

摘要

We propose Focused Page Rank (FPR) algorithm adaptation for the problem of scientific papers ranking. FPR is based on the Focused Surfer model, where the probability to follow the reference in a paper is proportional to its citation count. Evaluation on Citeseer autonomous digital library content showed that proposed model is a tradeoff between traditional citation count and basic Page Rank (PR). In contrast to basic Page Rank, proposed Focused Surfer model suffers less from the "outbound links" problem. We believe that FPR algorithm is closer to reality because highly cited papers are more visible and tend to attract more citations in future. This is in accordance with the one of the most significant principles of Scientometrics. No need for lexical analysis of the domain corpus and simplicity of implementation are among the strong points of the proposed model and make the proposed ranking technique attractive for acade-mia digital libraries.