Judge the Votes: A System to Classify Bug Reports and Give Suggestions

The aim of the research is to compare the modern reasoning approaches for the bug classification task with the previous state of the art models. We've used 4 language models. We tested them with zero-shot and few-shot settings. Few-shot examples are retrieved using RAG from the training set. We've selected the best model (o3-mini) and fed it the votes from ML models along with the retrieved bug reports with their labels.