Robot comment classifier
Summary
The article discusses a classifier for detecting whether source code comments were written by a robot or an LLM, including its methodology, features, and evaluation results. It explains why such detectors can be imperfect, outlines the data-labeling challenges, and highlights transfer limitations across models and domains.