Experiments on Code Clone Detection and Machine Learning
André Schäfer, Wolfram Amme, Thomas S. Heinze · 2022
The latest machine learning techniques are already capable of solving extremely complex problems and outperforming conventionally implemented software. Also in code clone detection, there is an increasing number of approaches that work with machine learning. In this paper, a machine learning model is trained with dominator trees generated from the bytecode of the BigCloneBench benchmark and similar performance results to previous work can be produced. However, further experiments using a rigorous approach with strict separation of data sets show that these previous results have little practical relevance.