GoodPoint: Learning Constructive Scientific Paper Feedback from Author Responses
Summary: arXiv:2604.11924v1 Announce Type: new
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.
Introduction
The integration of Large Language Models (LLMs) into scientific research processes is becoming increasingly vital. Researchers are exploring how these models can enhance the quality of scientific communication. GoodPoint represents a significant advancement in this area, focusing on the generation of constructive feedback that can support authors in refining their work.
Key Features of GoodPoint
- Author-Centric Feedback: GoodPoint emphasizes feedback that is both valid and actionable, ensuring that authors can implement suggestions effectively.
- Robust Dataset: The GoodPoint-ICLR dataset, comprising over 19,000 ICLR papers, provides a rich resource for training and evaluating the model.
- Advanced Training Techniques: The model employs fine-tuning on feedback deemed valid and actionable, along with preference optimization strategies.
- High Success Rate: Evaluation results indicate an 83.7% improvement in predicted success rates compared to base models.
- Expert Validation: Human expert evaluations confirm that GoodPoint enhances the practical value of feedback received by authors.
Implications for Researchers
The adoption of GoodPoint has significant implications for researchers across various scientific fields. By providing tailored feedback, the model not only aids in improving the quality of research but also enhances the overall scientific communication process. With the ability to surpass existing models in precision and effectiveness, GoodPoint is poised to become an essential tool in the research community.
Conclusion
GoodPoint represents a pivotal step forward in the utilization of LLMs for scientific research. By focusing on enhancing the feedback process, it empowers authors and encourages a more collaborative approach to research. As the landscape of scientific inquiry continues to evolve, tools like GoodPoint will be crucial in navigating the complexities of research communication.