Plagiarism detection in source code using machine learning

Abstract

The Source Code Plagiarism has become a major problem in today’s Educational World. The boost in the technology has led to the development of major IT sector and software industry. Thus, intellectual property of a being is not less important than any other valuable property. If the plagiarism reaches above the phase 6 it becomes almost impossible to detect it using the tools which are designed for the structural analysis. Therefore, we designed the new model for source code plagiarism detection, which uses the concepts of Machine Learning in order to fight with the higher phases of plagiarism. Conventional methods like structural methods, attribute counting method and graph-based analysis don’t produce the results with accuracy. Machine learning algorithms produce the most accurate result with continuous learning from the training modules. The three algorithms used are Naïve Bayes Algorithm, k-Nearest Neighbor and ADA Boost Meta Learning Algorithm. Since, no single algorithm can produce the result with accuracy, thus combining the algorithms help to produce results more accurately.