Weighted semantic plagiarism detection approach based on AHP decision model

The increasing rate of academic plagiarism is a social problem that engages institutions and publishers. Plagiarists try to mislead the plagiarism detection system using synonyms and inverted word order. Numerous algorithms tried to overcome these problems using structural and semantic detection. However, most of them focus on overcoming some challenges. Moreover, all of them consider the same significant degree for all terms of the documents. On the other hand, the time complexity is an essential parameter that must be considered. This paper presents an effective way to detect structural and semantic similarity degrees among two papers only using some part of the paper's content instead of all content, decreasing the time complexity. The similarity is calculated using a set of impressive terms and various combinations to augment plagiarism detection ability even if the word order is changed. Different weight is assigned to the word according to its position in various sections of the paper. Finally, an AHP (Analytical Hierarchy Process) model uses to calculate a weighted similarity. The results indicated that the proposed approach has more ability to detect semantic academic plagiarism, and the runtime is reduced compared to similar ones.