Plagiarism detection in arabic scripts using fuzzy information retreival

The nature of Arabic language structure exposes the need for fuzzy or vague concept to reveal dishonest practices in Arabic documents. In this paper, we present a statement-based plagiarism detection approach in Arabic scripts using fuzzy-set IR model. The degree of similarity is calculated and compared to a threshold value to judge whether two statements are the same or different. Our corpus collection has been built in which all stopwords were removed and non-stop words were stemmed for typical Arabic IR. The corpora have 100 documents with 4367 statements in total. Five query documents with about 250 plagiarized statements were constructed and tested. Experimental results show that fuzzyset IR successfully detected not only exact but also similar statements that have different structure. However, our Arabic fuzzy-set model approach does not handle the case of rewording with different synonyms/antonyms, a deficiency that will lead to future work of modeling the system using Arabic thesaurus. Keywordsfuzzy-set information retrieval; Arabic; plagiarism detection;