GoodPoint: Learning Constructive Scientific Paper Feedback from Author Responses

TL;DR

This paper introduces GoodPoint, a dataset and training method for large language models to generate constructive, actionable feedback for scientific papers, improving author guidance and research quality.

Contribution

The work creates a large annotated dataset of reviewer feedback and develops a fine-tuning approach that enhances LLMs to produce more valid and useful scientific paper feedback.

Findings

GoodPoint-trained models outperform baselines in feedback matching accuracy.

The approach improves success rate prediction by 83.7%.

Expert human evaluation confirms higher practical value of generated feedback.

Abstract

While LLMs hold significant potential to transform scientific research, we advocate for their use to augment and empower researchers rather than to automate research without human oversight. To this end, we study constructive feedback generation, the task of producing targeted, actionable feedback that helps authors improve both their research and its presentation. In this work, we operationalize the effectiveness of feedback along two author-centric axes-validity and author action. We first curate GoodPoint-ICLR, a dataset of 19K ICLR papers with reviewer feedback annotated along both dimensions using author responses. Building on this, we introduce GoodPoint, a training recipe that leverages success signals from author responses through fine-tuning on valid and actionable feedback, together with preference optimization on both real and synthetic preference pairs. Our evaluation on a benchmark of 1.2K ICLR papers shows that a GoodPoint-trained Qwen3-8B improves the predicted success rate by 83.7% over the base model and sets a new state-of-the-art among LLMs of similar size in feedback matching on a golden human feedback set, even surpassing Gemini-3-flash in precision. We further validate these findings through an expert human study, demonstrating that GoodPoint consistently delivers higher practical value as perceived by authors.