Lexical Generalisation for Word-level Matching in Plagiarism Detection

Proceedings of Recent Advances in Natural Language Processing, pages 704–709, Hissar, Bulgaria, 12-14 September 2011. Lexical Generalisation for Word-level Matching in Plagiarism Detection Miranda Chong University of Wolverhampton Miranda.Chong@wlv.ac.uk Lucia Specia University of Wolverhampton L.Specia@wlv.ac.uk Abstract Plagiarism has always been a concern in many sectors, particularly in education. With the sharp rise in the number of electronic resources available online, an increasing number of plagiarism cases has been observed in recent years. As the amount of source materials is vast, the use of plagiarism detection tools has become the norm to aid the investigation of possible plagiarism cases. This paper describes an approach to improve plagiarism detection by incorporating a lexical generalisation technique. The goal is to identify plagiarised texts even if they are paraphrased using different words. Experiments performed on a subset of the PAN‟10 corpus show that the matching approach involving lexical generalisation yields promising results, as compared to standard n-gram matching strategies. 1 Introduction Plagiarism is a growing challenge in modern society. In an attempt to maintain academic integrity, the use of plagiarism detection tools has become the norm in many higher education institutions. However, the methods used in these tools are mostly limited to comparisons of suspicious plagiarised texts and potential source texts at the string level. If the texts have not been copied verbatim, these tools are not able to identify the obfuscated texts effectively. Therefore, the accuracy of these methods is yet to reach a satisfactory level. This paper investigates the use of pre- processing, morphological and lexical semantics techniques from Natural Language Processing (NLP) in automatic plagiarism detection. The hypothesis is that by enhancing standard string matching approaches with linguistic information it is possible to improve the accuracy of plagiarism identification at the document level. More specifically, the goal is to generalise the text comparison to include morphological and lexical variations (synonyms). Different from previous work, instead of restricting the expansion of words in the documents to synonyms with the same sense, we use a simpler approach that considers all possible expansions. This approach does not require word sense disambiguation and is therefore less prone to common errors due to incorrect disambiguation. This paper is organised as follows: in Section 2 we describe related work in the plagiarism detection field using NLP; in Section 3 we outline the experimental settings; in Section 4 we present the results of our experiments; in Section 5 we discuss the findings; and in Section 6 we conclude and suggest future work. 2 Related Work Our focus is on external monolingual plagiarism detection of English documents. External detection refers to cases where potential source texts are available for comparison against suspicious plagiarised texts. Following the standard terminology in the field, we name suspicious document a potentially plagiarised text, and source document the possible origin of the plagiarised material. Current studies in this area have suggested the use of approaches such as n-gram matching between suspicious and potential source documents. NLP has only recently started to be exploited for …

Content truncated to 500 words