Design and Development of a Plagiarism Corpus in Thai for Plagiarism Detection

One of the main problems of creating a plagiarism corpus in Thai is that it is quite a difficult task to acquire the plagiarized documents with real cases due to the copyright issue. To solve the problem, we present a design and development of a Thai plagiarism corpus to evaluate and compare plagiarism detection algorithms for Thai. The corpus is developed by using the simulated plagiarism method based on Thai Wikipedia articles and web page articles. For this method, we provide a Thai plagiarism annotation tool and a Thai plagiarism guideline for assisting human annotators to plagiarize text passages. Our corpus contains simulated cases of plagiarized documents based on four classes of Thai plagiarism and linguistic mechanisms including copy-based change, lexicon-based change, structure-based change, and semantic-based change. We show that the suspicious documents in the corpus are manually created by using different obfuscation strategies, which make the suspicious documents more realistic and challenging. We then believe that the corpus developed in this paper will be a valuable contribution in the development, comparison, and evaluation of plagiarism detection algorithms. Moreover, our corpus is free and publicly available for research purposes.