A New Algorithm of Grouping Cockroaches Classifier (GCC) for Textual Plagiarism Detection

Abstract

In the last decade with the new technology, it is important to allow users to access information freely, while at the same time, restrict them from illegal copying and distribution of information. In the age of information technologies plagiarism has become a topical subject in the digital world and turned into a serious problem. The author's work deals with the development of a new system for combating this phenomenon using a new insect behaviour algorithm called Groping cockroaches classifier GCC. Each suspicious text (cockroach) will be classified (hidden) in a class (shelter) that can be plagiarism or no-plagiarism, using a security function that is based on the attractiveness of each class (calculated using the aggregation operators (shelter darkness, congeners attraction and security quality)) and the displacement probability (calculated using the naive Bayes algorithm). The experimental results performed on the Pan 09 dataset and using the validation measures (recall, precision, f-measure, and entropy), have demonstrated that GCC has clear advantages over others plagiarism detection techniques existed in literature. Finally, a set of service was added in order to detect the different cases of plagiarism such as plagiarism with translation, plagiarism of idea, plagiarism with synonymy, and plagiarism paraphrase.