Plagiarism Detection in Arabic Documents using word2vector and Arabic WordNet

Abstract

Plagiarism detection has become a latest research area in Natural Language Processing Field. In today's with the huge available content of Arabic articles on the internet this make the text plagiarism is so easy and Spread widely in academic society, so to decrease and prevent this Harmful habit many algorithms has been developed to detect plagiarized texts in many famous languages and in Arabic language, in this article we have developed an algorithm to detect plagiarism in Arabic text, in this algorithm we have used word2vector technology which transforms texts to numeric vectors and it keep the syntax and the meaning of sentences, and we have used Arabic WordNet to overcome of the limitation of word2vector models, we have used Twitter Continuous Bag-of-words word2vector model which built from huge repository of Arabic tweets in different topics, and using of Arabic WordNet is to finds the numeric vectors of non- existing words in word2vector model so in this case we try to find numeric vectors for one synonym word of the non-existing words in the model