Authorship analysis: identifying the author of a program

Abstract

Authorship analysis on computer software is a difficult problem. In this paper we explore the classification of programmers' style, and try to find a set of characteristics that remain constant for a significant portion of the programs that this programmer might produce. Our goal is to show that it is possible to identify the author of a program by examining programming style characteristics. Within a closed environment, the results of this paper support the conclusion that, for a specific set of programmers, it is possible to identify the author of any individual program. Also, based on previous work and our observations during the experiments described herein, we believe that the probability of finding two programmers who share exactly those same characteristics should be very small.