Optimization Methods Applied to Plagiarism Detection in Programming Exercises in C Language

This work presents an extension of a plagiarism detection model in programming exercises based on graphs. The original model, which uses the solution of the Prize-Collecting Shortest Path Problem with Maximum Prize (PMCPM) in a complete bipartite graph to quantify the similarity between Python codes, was adapted to support the C language. To solve the PMCPM, two methods are proposed: an exact one, based on Dynamic Programming, and a heuristic one, combining the Multi-Start Metaheuristic with a local search based on the Variable Neighborhood Descent (VND) strategy. Additionally, a benchmark dataset based on real programming exercises was constructed to validate the model and the proposed methods. The computational experiments demonstrated that the model is capable of consistently representing the similarities between C language source codes, paving the way for future research.