Measurements of program similarity in identical task environments

Abstract

This paper summarizes the results of a study which compared the efficiency of two methods of measuring program similarity in the context of novice programmers trying to reach identical objectives. Both methods look for similarity by comparing 'program profiles'. Such profiles are created by feature extraction routines which map each program onto a tuple <f 1 , f 2 ,..., f n > where each f i is a count of an occurrence of a particular feature. A comparison routine is then invoked which detects similarities between tuples. The results showed that in this environment the comparison routine based on the Halstead metric failed to perform as well as a conceptually simpler alternative.