Plagiarism Detection Using the Levenshtein Distance and Smith-Waterman Algorithm

Abstract

Plagiarism in texts is issues of increasing concern to the academic community. Now most common text plagiarism occurs by making a variety of minor alterations that include the insertion, deletion, or substitution of words. Such simple changes, however, require excessive string comparisons. In this paper, we present a hybrid plagiarism detection method. We investigate the use of a diagonal line, which is derived from Levenshtein distance, and simplified Smith-Waterman algorithm that is a classical tool in the identification and quantification of local similarities in biological sequences, with a view to the application in the plagiarism detection. Our approach avoids globally involved string comparisons and considers psychological factors, which can yield significant speed-up by experiment results. Based on the results, we indicate the practicality of such improvement using Levenshtein distance and Smith-Waterman algorithm and to illustrate the efficiency gains. In the future, it would be interesting to explore appropriate heuristics in the area of text comparison.