A Program Plagiarism Detection Approach Based On Abstract Syntax Tree

It is known that difficulties generated by plagiarism in students programming assignment submissions harass instructors. Therefore, efficient detection of the program similarity shows its importance to the educational procedure. In this paper, a plagiarism detection approach based on the program's grammar architecture is introduced, which can evaluate the similarity by comparing the units generated from the programs' syntax trees. Experimental study demonstrates that this approach is robust to many plagiarism strategies. Furthermore, system based on the approach is proposed to detect whether plagiarism exists or not in program sets through a clustering algorithm. Tested by a set of submissions, the comparative evaluation shows that our system per-forms better than Jplag, a famous plagiarism detection system.