Program Plagiarism Detection with Dynamic Structure

• Main Authors: Sheng Chuan Hsu, 許聖泉
• Other Authors: Jong Yih Kuo
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/b3ksvu

碩士 === 國立臺北科技大學 === 資訊工程系研究所 === 105 === In this study, we develop an automated program alignment system: SPDS, which allows users to determine whether there is any possibility of plagiarism between them. In the function structural part, we use Dynamic Slicing to analyze the behavior of the function, combine the Combinatorial Decision Coverage to generate valid inputs for Dynamic Slicing, and convert the results to Program Dependence Graph to calculate structural similarity; and use Def-Use to analyze the structure of the function The attribute similarity is calculated by using the Cosine Similarity, however, if the attributes belongs to the users own category, the recursive method is used for the calculation. After the comment filtering out non-related parts, it applies Winnowing algorithm for comment similarity calculation. Finally, according to the structure of the program, the weights of each similarity are set. Based on the characteristics of the current program structure, the system searches the past case to provide the available weights. And the weights of each users feedback will also be stored for future reference.