Supplementary Material

This replication package contains supplementary material for the paper "MERCAnnot: Automated Annotation of Modification-Eliciting Code Review Comments with Large Language Models". The package is organized as follows:

• Data/ The datasets used in this paper
  • Data/CodeReviewerTrainDataset.xlsx The original desiredness score results of the training dataset analysis for codereviewer.
  • Data/CodeReviewerTestDataset.xlsx The original desiredness score results of the testing dataset analysis for codereviewer.
  • Data/OriginFinetuneDataset.jsonl The training set for the automatic review comment generation task of LLaMA-Reviewer.
  • Data/DesiredFinetuneDataset.jsonl The training set for the automatic review comment generation task of Desiview4FT.
  • Data/AlignmentDataset.jsonl The training set for the automatic review comment generation task of Desiview4FA.
  • Data/TestDataset.jsonl The testing set for the automatic review comment generation task.
• Scorer/ Source code for scoring desiredness code fixes.
• Finetune/ The fine-tuning code for the review comment generation task.
• Alignment/ The alignment code for the review comment generation task.