Ontological approach to detect plagiarism between Arabic documents based on semantic similarity

Abstract

In the context of Arabic plagiarism detection systems (APDS) using an Arabic ontology, and permitting these systems to support semantic representation, to better verify the originality of the research and meet the needs of researchers, this thesis aims to semantically index documents by selecting the best elements from Arabic WordNet. We have analyzed each sentence in the document, have extracted Part-of-Speech (PoS), have extracted Arabic WordNet synonyms for each word and then we have created the first index. The latter is then used by the Lucene program to create the second index, which is used to detect plagiarism in Arabic documents. Our experience is based on a corpus of Arabic text, which we have manually created with the interaction of an expert. The results obtained showed us that the proposed system has proved its performance and its efficacy in detecting the plagiarism in Arabic documents. key words : Arabic Documents, Arabic Ontology, Semantic Indexing, Semantic Similarity, Plagiarism Detection.