Computer Program Plagiarism Detection: The Limits of the Halstead Metric

Abstract

Plagiarism detection is an increasingly important issue in educational environments. This article employs two alternative metrics in the identification of program similarity. The first is the Halstead metric drawn from the discipline of software science. The second is an ad hoc metric drawn from program grading experience, and identified by means of factor analysis. The ad hoc metric proves to be more useful in identical-task environments. Possible explanations for, as well as some larger implications of, this result are considered.