Machine learning tool and meta-heuristic based on genetic algorithms for plagiarism detection over mail service

Abstract

One of the most modern problems that computer science try to resolve is the plagiarism, in this article we present a new approach for automatic plagiarism detection in world of mail service. Our system is based on the n-gram character for the representation of the texts and tfidf as weighting to calculate the importance of term in the corpus, we use also a combination between the machine learning methods as a way to detect if a document is plagiarized or not, we use pan 09 corpus for the construction and evaluation of the prediction model then we simulate a meta-heuristic method based on genetic algorithms with a variations of parameters to know if it can improve the results. The main objective of our work is to protect intellectual property and improve the efficiency of plagiarism detection system.