Detection of Mechanisms for Unfair Increase of Scientometric Indexes

Abstract

The goal of the research is to detect groups of authors that increase their scientometric indexes through excessively frequent mutual citations. In a citation graph with vertices and edges representing authors and «author-cites-author» relations respectively, suspicious groups might be cliques that have few citations from other authors and, conversely, many internal citations. Metadata on scientific publications from the Crossref service was retrieved, filtered and reduced to an easy-to-process form, and then uploaded to the local MySQL database. After the data transfer phase the undirected citation graph and its connectivity components were extracted, the maximum cliques were found and their statistical characteristics were analyzed. The most suspicious collectives were detected using these characteristics. This method can be used to detect groups of authors who cite each other by agreement.