А method for detecting implicit plagiarism in scientific and technical texts

The paper considers the process of automatic detection of implicit plagiarism in documents on the base of comparison of their formalized representations and calculation of the measures of local semantic similarity of concepts and global semantic similarity of text fragments. In solving this problem we developed a model of the semantic structure of texts and methods for formalization and detection of semantic proximity of the texts under comparison. We also developed the methods for identification of the text fragments similar in semantic structure. The main advantage of this method is that it makes it possible to detect different kinds of plagiarism including the most complex cases of implicit plagiarism. In the study, the results of the work were compared to the results obtained with the use of the method of “shingles”. The proposed method showed high efficiency.