---
title: "Feedback Is The Key for Automated Survey Generation — Xu 2026"
description: "The framework can build a survey from a topic or improve an existing draft, and it optionally accepts human feedback. 23 references."
authors: ["Tianyi Xu", "Zhe Zhao", "Tianshuo Wei", "Yiqun Kou", "Liuliu Han", "Ye Wei"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.1904/"
doi: "10.18653/v1/2026.findings-acl.1904"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/xu-2026-feedback-is-key-automated-survey/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/xu-2026-feedback-is-key-automated-survey/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/33/) / Feedback Is The Key for Automated Survey Generation

# Feedback Is The Key for Automated Survey Generation

**Authors:** [Tianyi Xu](https://md.research.elamur.ai/free/en/authors/tianyi-xu/), [Zhe Zhao](https://md.research.elamur.ai/free/en/authors/zhe-zhao/), [Tianshuo Wei](https://md.research.elamur.ai/free/en/authors/tianshuo-wei/), Yiqun Kou, Liuliu Han and [Ye Wei](https://md.research.elamur.ai/free/en/authors/ye-wei/) · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.1904/) · [PDF](https://aclanthology.org/2026.findings-acl.1904.pdf) · [DOI 10.18653/v1/2026.findings-acl.1904](https://doi.org/10.18653/v1/2026.findings-acl.1904) · [OpenAlex](https://openalex.org/W7166855891)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
LLM-based survey writers such as AutoSurvey, SurveyX and SurveyForge usually draft in one forward pass. Their output therefore often contains hallucinated claims, weak citations and loose structure. FIKSurvey adds the review-and-revise cycle that human authors rely on. A Writer LLM produces the text, and a separate Helper LLM acts as a peer reviewer. The reviewer gives feedback at three levels: the outline (primary, then secondary), the content (depth and readability) and the citations.

To check citations, the system extracts claim–citation pairs and runs a sliding-window NLI check against the cited papers. If a claim is unsupported, it retrieves top-K alternative papers and either swaps the reference or rewrites the claim. A revision is kept only when it raises a quantitative rubric score.

The framework can build a survey from a topic or improve an existing draft, and it optionally accepts human feedback. Ablations show that feedback, especially citation feedback, sharply raises citation recall and precision. The work matters because it treats structured, verifiable critique as more important for trustworthy automated academic writing than raw generation power.

## Key points
- The pipeline has four stages: candidate paper preparation, outline generation, drafting and improvement.
  - Papers are first sorted into four types: method, theory, benchmark or survey.
  - Each paper is then turned into a JSON attribute tree, an approach that follows SurveyX.
- The feedback design is supported by an analysis of 183 survey reviews collected from OpenReview.
- Ablation on 10 randomly sampled topics:
  - The full system reaches about 97.2% citation recall and 93.4% citation precision.
  - Removing citation feedback alone drops these to about 80.5% and 65.7%.
- Each feedback type plays a separate role:
  - Removing outline feedback mainly lowers the structure score (5.00 → 4.40), while citation quality stays roughly unchanged.
  - Removing all three feedback types lowers average content quality from 4.97 to 4.83.
- The authors list two limitations:
  - Rubric criteria currently carry equal weight, so adaptive weighting by topic or venue is proposed as future work.
  - Uncertainty-aware feedback could help in cases where reviewer critiques and LLM judgments diverge.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Huazhong University of Science and Technology](https://md.research.elamur.ai/free/en/universities/huazhong-university-of-science-and-technology/) — University, Wuhan, China · 59 works · 8,310 citations
- [City University of Hong Kong](https://md.research.elamur.ai/free/en/universities/city-university-of-hong-kong/) — University, Hong Kong, China · 59 works · 3,480 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🗄️ Information Systems](https://md.research.elamur.ai/free/en/shelves/computer-science/information-systems/) · 1,730 works · [🎭 History and Philosophy of Science](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/history-and-philosophy-of-science/) · 62 works

## Cite

Cite as: Tianyi Xu, Zhe Zhao, Tianshuo Wei, Yiqun Kou, Liuliu Han and Ye Wei. 2026. Feedback Is The Key for Automated Survey Generation. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{xu2026feedback,
  title = {{Feedback Is The Key for Automated Survey Generation}},
  author = {Tianyi Xu, Zhe Zhao, Tianshuo Wei, Yiqun Kou, Liuliu Han and Ye Wei},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1904},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1904}
}
```

- [← previous FedReFT: Federated Representation Fine-Tuning with All-But-Me Aggregation](https://md.research.elamur.ai/free/en/papers/siddika-2026-fedreft-federated-representation-fine-tuning/)
- [next → Feedback-Aware Prompt Optimization Framework for Generating Job Postings](https://md.research.elamur.ai/free/en/papers/maharjan-2026-feedback-aware-prompt-optimization-framework/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/33/)

## References

23 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Lutz Bornmann, Robin Haunschild, and Rüdiger Mutz. 2021. Growth rates of modern science: a latent piecewise growth curve approach to model publication numbers from established and new literature databases. Humanities and Social Sciences Communications, 8(1):1–15. [find](https://scholar.google.com/scholar?q=Lutz+Bornmann%2C+Robin+Haunschild%2C+and+R%C3%BCdiger+Mutz%2E+2021%2E+Growth+rates+of+modern+science%3A+a+latent+piecewise+growth+curve+approach+to+model+publication+numbers+from+established+and+new+literature+datab)
2. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, and 1 others. 2020. [Language models are few-shot learners](https://md.research.elamur.ai/free/en/papers/brown-2020-language-models-few-shot-learners/). Advances in neural information processing systems, 33:1877–1901.
3. Jingqiang Chen and Hai Zhuge. 2019. Automatic generation of related work through summarizing citations. Concurrency and Computation: Practice and Experience, 31(3):e4261. [find](https://scholar.google.com/scholar?q=Jingqiang+Chen+and+Hai+Zhuge%2E+2019%2E+Automatic+generation+of+related+work+through+summarizing+citations%2E+Concurrency+and+Computation%3A+Practice+and+Experience%2C+31%283%29%3Ae4261%2E)
4. Juhwan Choi, JungMin Yun, Changhun Kim, and YoungBin Kim. 2025. Position paper: How should we responsibly adopt llms in the peer review process? Submitted to ACL Rolling Review-July, 2025. [find](https://scholar.google.com/scholar?q=Juhwan+Choi%2C+JungMin+Yun%2C+Changhun+Kim%2C+and+YoungBin+Kim%2E+2025%2E+Position+paper%3A+How+should+we+responsibly+adopt+llms+in+the+peer+review+process%3F+Submitted+to+ACL+Rolling+Review%2DJuly%2C+2025%2E)
5. Shehzaad Dhuliawala, Mojtaba Komeili, Jing Xu, Roberta Raileanu, Xian Li, Asli Celikyilmaz, and Jason Weston. 2024. [Chain-of-verification reduces hallucination in large language models](https://md.research.elamur.ai/free/en/papers/dhuliawala-2023-chain-verification-reduces-hallucination-large/). In Findings of the association for computational linguistics: ACL 2024, pages 3563–3578.
6. Michael Fire and Carlos Guestrin. 2019. Overoptimization of academic publishing metrics: observing goodhart’s law in action. GigaScience, 8(6):giz053. [find](https://scholar.google.com/scholar?q=Michael+Fire+and+Carlos+Guestrin%2E+2019%2E+Overoptimization+of+academic+publishing+metrics%3A+observing+goodhart%E2%80%99s+law+in+action%2E+GigaScience%2C+8%286%29%3Agiz053%2E)
7. Cong Duy Vu Hoang and Min-Yen Kan. 2010. Towards automated related work summarization. In Coling 2010: Posters, pages 427–435. [find](https://scholar.google.com/scholar?q=Cong+Duy+Vu+Hoang+and+Min%2DYen+Kan%2E+2010%2E+Towards+automated+related+work+summarization%2E+In+Coling+2010%3A+Posters%2C+pages+427%E2%80%93435%2E)
8. Yue Hu and Xiaojun Wan. 2014. Automatic generation of related work sections in scientific papers: an optimization approach. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1624–1633. [find](https://scholar.google.com/scholar?q=Yue+Hu+and+Xiaojun+Wan%2E+2014%2E+Automatic+generation+of+related+work+sections+in+scientific+papers%3A+an+optimization+approach%2E+In+Proceedings+of+the+2014+Conference+on+Empirical+Methods+in+Natural+Langua)
9. Ryo Kamoi, Yusen Zhang, Nan Zhang, Jiawei Han, and Rui Zhang. 2024. [When can llms actually correct their own mistakes? a critical survey of self-correction of llms](https://md.research.elamur.ai/free/en/papers/kamoi-2024-when-can-llms-actually-correct/). Transactions of the Association for Computational Linguistics, 12:1417–1440.
10. Xun Liang, Jiawei Yang, Yezhaohui Wang, Chen Tang, Zifan Zheng, Shichao Song, Zehao Lin, Yebin Yang, Simin Niu, Hanyu Wang, and 1 others. 2025. Surveyx: Academic survey automation via large language models. arXiv preprint arXiv:2502.14776. — [arXiv:2502.14776](https://arxiv.org/abs/2502.14776) [find](https://scholar.google.com/scholar?q=Xun+Liang%2C+Jiawei+Yang%2C+Yezhaohui+Wang%2C+Chen+Tang%2C+Zifan+Zheng%2C+Shichao+Song%2C+Zehao+Lin%2C+Yebin+Yang%2C+Simin+Niu%2C+Hanyu+Wang%2C+and+1+others%2E+2025%2E+Surveyx%3A+Academic+survey+automation+via+large+language+m)
11. Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, and 1 others. 2023. Self-refine: Iterative refinement with self-feedback. Advances in Neural Information Processing Systems, 36:46534–46594. [find](https://scholar.google.com/scholar?q=Aman+Madaan%2C+Niket+Tandon%2C+Prakhar+Gupta%2C+Skyler+Hallinan%2C+Luyu+Gao%2C+Sarah+Wiegreffe%2C+Uri+Alon%2C+Nouha+Dziri%2C+Shrimai+Prabhumoye%2C+Yiming+Yang%2C+and+1+others%2E+2023%2E+Self%2Drefine%3A+Iterative+refinement+with)
12. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2020. [Exploring the limits of transfer learning with a unified text-to-text transformer](https://md.research.elamur.ai/free/en/papers/raffel-2019-exploring-limits-transfer-learning-unified/). Journal of machine learning research, 21(140):1–67.
13. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. [Reflexion: Language agents with verbal reinforcement learning](https://md.research.elamur.ai/free/en/papers/shinn-2023-reflexion-language-agents-verbal-reinforcement/). Advances in Neural Information Processing Systems, 36:8634–8652.
14. Weihang Su, Anzhe Xie, Qingyao Ai, Jianming Long, Jiaxin Mao, Ziyi Ye, and Yiqun Liu. 2025. Surge: A benchmark and evaluation framework for scientific survey generation. arXiv preprint arXiv:2508.15658. — [arXiv:2508.15658](https://arxiv.org/abs/2508.15658) [find](https://scholar.google.com/scholar?q=Weihang+Su%2C+Anzhe+Xie%2C+Qingyao+Ai%2C+Jianming+Long%2C+Jiaxin+Mao%2C+Ziyi+Ye%2C+and+Yiqun+Liu%2E+2025%2E+Surge%3A+A+benchmark+and+evaluation+framework+for+scientific+survey+generation%2E+arXiv+preprint+arXiv%3A2508%2E1565)
15. Xiaoping Sun and Hai Zhuge. 2019. Automatic generation of survey paper based on template tree. In 2019 15th International Conference on Semantics, Knowledge and Grids (SKG), pages 89–96. IEEE. [find](https://scholar.google.com/scholar?q=Xiaoping+Sun+and+Hai+Zhuge%2E+2019%2E+Automatic+generation+of+survey+paper+based+on+template+tree%2E+In+2019+15th+International+Conference+on+Semantics%2C+Knowledge+and+Grids+%28SKG%29%2C+pages+89%E2%80%9396%2E+IEEE%2E)
16. Yidong Wang, Qi Guo, Wenjin Yao, Hongbo Zhang, Xin Zhang, Zhen Wu, Meishan Zhang, Xinyu Dai, Qingsong Wen, Wei Ye, and 1 others. 2024a. [Autosurvey: Large language models can automatically write surveys](https://md.research.elamur.ai/free/en/papers/wang-2024-autosurvey-large-language-models-automatically/). Advances in neural information processing systems, 37:115119–115145.
17. Yidong Wang, Qi Guo, Wenjin Yao, Hongbo Zhang, Xin Zhang, Zhen Wu, Meishan Zhang, Xinyu Dai, Min Zhang, Qingsong Wen, Wei Ye, Shikun Zhang, and Yue Zhang. 2024b. [Autosurvey: Large language models can automatically write surveys](https://md.research.elamur.ai/free/en/papers/wang-2024-autosurvey-large-language-models-automatically/). In The Thirty-eighth Annual Conference on Neural Information Processing…
18. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, and 1 others. 2022. [Chain-of-thought prompting elicits reasoning in large language models](https://md.research.elamur.ai/free/en/papers/wei-2022-chain-thought-prompting-elicits-reasoning/). Advances in neural information processing systems, 35:24824– 24837.
19. Xiangchao Yan, Shiyang Feng, Jiakang Yuan, Renqiu Xia, Bin Wang, Lei Bai, and Bo Zhang. 2025. [Surveyforge: On the outline heuristics, memory-driven generation, and multi-dimensional evaluation for automated survey writing](https://md.research.elamur.ai/free/en/papers/yan-2025-surveyforge-outline-heuristics-memory-driven/). In Proceedings of the 63rd Annual Meeting of the Association for Computationa…
20. Itai Yanai and Martin J. Lercher. 2024. It takes two to think. Nature Biotechnology, 42(1):18–19. [find](https://scholar.google.com/scholar?q=Itai+Yanai+and+Martin+J%2E+Lercher%2E+2024%2E+It+takes+two+to+think%2E+Nature+Biotechnology%2C+42%281%29%3A18%E2%80%9319%2E)
21. Zhe Yang, Yichang Zhang, Yudong Wang, Ziyao Xu, Junyang Lin, and Zhifang Sui. 2025. Confidence vs critique: A decomposition of self-correction capability for llms. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3998–4014. [find](https://scholar.google.com/scholar?q=Zhe+Yang%2C+Yichang+Zhang%2C+Yudong+Wang%2C+Ziyao+Xu%2C+Junyang+Lin%2C+and+Zhifang+Sui%2E+2025%2E+Confidence+vs+critique%3A+A+decomposition+of+self%2Dcorrection+capability+for+llms%2E+In+Proceedings+of+the+63rd+Annual+Me)
22. Jiebin Zhang, J Yu Eugene, Qinyu Chen, Chenhao Xiong, Dawei Zhu, Han Qian, Mingbo Song, Xiaoguang Li, Qun Liu, and Sujian Li. 2024. Retrieval-based full-length wikipedia generation for emergent events. Preprint. [find](https://scholar.google.com/scholar?q=Jiebin+Zhang%2C+J+Yu+Eugene%2C+Qinyu+Chen%2C+Chenhao+Xiong%2C+Dawei+Zhu%2C+Han+Qian%2C+Mingbo+Song%2C+Xiaoguang+Li%2C+Qun+Liu%2C+and+Sujian+Li%2E+2024%2E+Retrieval%2Dbased+full%2Dlength+wikipedia+generation+for+emergent+events)
23. Ziyan Zhang, Yang Hou, Chen Gong, and Zhenghua Li. 2025. Self-correction makes LLMs better parsers. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 6749–6762, Suzhou, China. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Ziyan+Zhang%2C+Yang+Hou%2C+Chen+Gong%2C+and+Zhenghua+Li%2E+2025%2E+Self%2Dcorrection+makes+LLMs+better+parsers%2E+In+Findings+of+the+Association+for+Computational+Linguistics%3A+EMNLP+2025%2C+pages+6749%E2%80%936762%2C+Suzhou%2C+C)

*Updated 2 November 2026*