Mossad: defeating software plagiarism detection

Abstract

Automatic software plagiarism detection tools are widely used in educational settings to ensure that submitted work was not copied. These tools have grown in use together with the rise in enrollments in computer science programs and the widespread availability of code on-line. Educators rely on the robustness of plagiarism detection tools; the working assumption is that the effort required to evade detection is as high as that required to actually do the assigned work. This paper shows this is not the case. It presents an entirely automatic program transformation approach, MOSSAD, that defeats popular software plagiarism detection tools. MOSSAD comprises a framework that couples techniques inspired by genetic programming with domain-specific knowledge to effectively undermine plagiarism detectors. MOSSAD is effective at defeating four plagiarism detectors, including Moss and JPlag. MOSSAD is both fast and effective: it can, in minutes, generate modified versions of programs that are likely to escape detection. More insidiously, because of its non-deterministic approach, MOSSAD can, from a single program, generate dozens of variants, which are classified as no more suspicious than legitimate assignments. A detailed study of MOSSAD across a corpus of real student assignments demonstrates its efficacy at evading detection. A user study shows that graduate student assistants consistently rate MOSSAD-generated code as just as readable as authentic student code. This work motivates the need for both research on more robust plagiarism detection tools and greater integration of naturally plagiarism-resistant methodologies like code review into computer science education.

Contract cheating: an increasing challenge for global academic community arising from COVID-19

“…Universities use software systems such as Turnitin, SafeAssign, SNITCH, Cross-Check, AntiPlag and other related software to detect plagiarism, and these systems are constantly improving (Pàmieset al, 2020;Bolkan, 2006;Jiffriya et al, 2013;Niezgoda & Way, 2006;Walker, 2010;Zhang, 2010). However, currently, there is no fully reliable tool with which to detect all cases of plagiarism and contract cheating, and some researchers have developed systems to defeat plagiarism detection (Devore-McDonald & Berger, 2020). Existing systems have gaps that are being addressed but are not resolved at this stage (Bin-Habtoor & Zaher, 2012).…”

Detecting Plagiarism as Out-of-distribution Samples for Large-scale Programming Contests

“…We test our method against the Mossad approach 4 to plagiarism detection evasion (Devore-McDonald and Berger, 2020). After applying our method, the rank of the pair of the original and the mutation are consistently below 10 in the 5 groups tested.…”

“…Suspicious code pair filtering. For the convenient and easily interpretable quantification of the possibility of plagiarism, many existing methods employ a single suspiciousness index for each source code pair (Devore-McDonald and Berger, 2020;Freire et al, 2007;Ajmal et al, 2013;Yasaswi et al, 2017;Sulistiani and Karnalim, 2019; R. Wu, A. Lv, Q. Zhao 94 Jiffriya et al, 2014). With the aid of the suspicious index, human operators are able to filter the most relevant pairs for manual inspection (Devore-McDonald and Berger, 2020;Freire et al, 2007).…”

“…For the convenient and easily interpretable quantification of the possibility of plagiarism, many existing methods employ a single suspiciousness index for each source code pair (Devore-McDonald and Berger, 2020;Freire et al, 2007;Ajmal et al, 2013;Yasaswi et al, 2017;Sulistiani and Karnalim, 2019; R. Wu, A. Lv, Q. Zhao 94 Jiffriya et al, 2014). With the aid of the suspicious index, human operators are able to filter the most relevant pairs for manual inspection (Devore-McDonald and Berger, 2020;Freire et al, 2007). The suspiciousness indices are typically computed from the feature vectors of the individual source codes using vector similarity functions, such as the Euclidean distance (Ajmal et al, 2013;Yasaswi et al, 2017) and cosine similarity (Rahutomo et al, 2012;Sulistiani and Karnalim, 2019;Jiffriya et al, 2014).…”

Similarity Identification Based on Word Trigrams Using Exact String Matching Algorithms

“…Meanwhile, in the Winnowing algorithm, fingerprints are determined by first grouping all hash values into windows based on a certain width. The rightmost minimum hash value obtained from each window is used as fingerprints [6], [7], [8], [9]. To prevent collisions in hash values, the Rolling Hash formula is used with the base number in the procedure using prime numbers [4], [10], [11].…”