UPPC - Urdu Paraphrase Plagiarism Corpus

UPPC - Urdu Paraphrase Plagiarism Corpus Muhammad Sharjeel∗†, Paul Rayson∗, Rao Muhammad Adeel Nawab† ∗School of Computing and Communications, Lancaster University, UK †Department of Computer Science, COMSATS Institute of Information Technology, Lahore, Pakistan Email: {s.muhammad6,p.rayson}@lancaster.ac.uk, adeelnawab@ciitlahore.edu.pk Abstract Paraphrase plagiarism is a significant and widespread problem and research shows that it is hard to detect. Several methods and automatic systems have been proposed to deal with it. However, evaluation and comparison of such solutions is not possible because of the unavailability of benchmark corpora with manual examples of paraphrase plagiarism. To deal with this issue, we present the novel development of a paraphrase plagiarism corpus containing simulated (manually created) examples in the Urdu language - a language widely spoken around the world. This resource is the first of its kind developed for the Urdu language and we believe that it will be a valuable contribution to the evaluation of paraphrase plagiarism detection systems. Keywords: Paraphrase Plagiarism, Corpus Generation, Urdu Plagiarism Detection, Natural Language Processing 1. Introduction Plagiarism is the use of material without the specification of its source. The freely available nature of on-line text has made unacknowledged reuse much more prevalent. Pla- giarism has long been considered to be a serious academic offence (Ali et al., 2011). Surveys conducted in the past reveal that more than 90% of students are involved in pla- giarism (Butakov and Scherbinin, 2009) while 40% com- mitted plagiarism during assignment submission (Osman et al., 2012). These figures are alarmingly high. Con- sequently, detection of plagiarism has attracted attention of the research community (McCabe et al., 2006; Chong et al., 2010; Potthast et al., 2010b; Sanchez-Perez et al., 2014) and higher education institutions are now regularly using automatic plagiarism detection software(s) to check student’s work for plagiarism. The task of identifying plagiarism from a suspicious doc- ument (i.e. one that is suspected to contain plagiarised text) can be accomplished using: (1) intrinsic plagiarism detection; whether the entire document is written by one single author or not and (2) extrinsic plagiarism detection; identification of the original document(s) which were used in producing the plagiarised one. Detection of plagiarism source(s) is difficult because there can be different levels of plagiarism: (1) word-to-word plagiarism, where the text is reused verbatim from the source, (2) paraphrase plagia- rism, when the content of the text is obfuscated using differ- ent linguistic techniques and (3) plagiarism of idea, when the main idea of the original document is reused indepen- dently of the words in the original text (Martin, 1994). Re- search indicates that the current plagiarism detection sys- tems can only detect verbatim copies and paraphrase pla- giarism cases are hard to detect and therefore, are an open challenge (Maurer et al., 2006; Potthast et al., 2010a; Pot- thast et al., 2011; Weber-Wulff et al., 2013). To develop and evaluate paraphrase plagiarism detection systems, we need benchmark corpora with examples that imitate real life cases. Benchmark datasets have been de- veloped previously (Potthast et al., 2010b; Clough and Stevenson, 2011) but majority …

Content truncated to 500 words