Separating the articles of authors with the same name

Abstract

I describe a method to separate the articles of different authors with the same name. It is based on a distance between any two publications, defined in terms of the probability that they would have as many coincidences if they were drawn at random from all published documents. Articles with a given author name are then clustered according to their distance, so that all articles in a cluster belong very likely to the same author. The method has proven very useful in generating groups of papers that are then selected manually. This simplifies considerably citation analysis when the author publication lists are not available.