Elamur Research⌕
☰
home / papers / 2026 / Think Before you Write: QA-Guided Reasoning for Character Descriptions in Books

Think Before you Write: QA-Guided Reasoning for Character Descriptions in Books

Paper · 2026 · Argyrios Papoudakis, Mirella Lapata and Frank Keller

Summary

Writing descriptions of characters from full-length books is difficult. A model must follow how a character's relationships and circumstances change, combine clues spread across more than 100k tokens, and infer details the text only implies. The authors find that reasoning-capable LLMs perform worse on this task when their native thinking is enabled than when the thinking trace is left empty. They therefore split the pipeline into two parts. A reasoning model (R_φ) produces a structured trace of question–answer pairs about the character. A generation model (G_θ) then receives this trace inside its thinking tokens and writes the description. The reasoner is trained with GRPO against gold descriptions and silver QA traces derived from them, which avoids designing rewards for open-ended text. The generator is either fine-tuned with SFT or used zero-shot. The method can be layered on long-context LLMs or on chunk-based strategies such as hierarchical merging, incremental updating and retrieval. On BookWorm and CroSS it improves faithfulness, informativeness and grounding over strong baselines. This offers a workable recipe for RL-trained reasoning in tasks whose outputs cannot be automatically verified.

Key points

  • Native LLM reasoning hurts grounding. On CroSS with Qwen-3-8B, NLI for Coref-32k falls from 50.01 to 46.84, and EntMent for Lead-128k falls from 35.48 to 31.18.
  • The datasets are BookWorm (Project Gutenberg books, mostly novels and plays) and CroSS (recently published novels). Models tuned on BookWorm transfer to CroSS without further training.
  • On CroSS with Lead-128k, guided-QA using GRPO traces scores best among Qwen-3-8B setups (PRISMA 23.78, QA F1 14.42). GPT-4.1-mini with full context remains an upper bound (PRISMA 28.49, NLI 72.67). However, it trails the guided Qwen-3-8B on entity coverage (EntMent 28.49 vs. 36.84, as reported).
  • Trace quality drives the gains, not trace length. GRPO-optimized traces work best, and oracle traces far exceed all automatic ones, which leaves clear room for improvement.
  • Evaluation uses PRISMA, QA F1, NLI entailment, entity-mention F1 and Rouge-L. Code and data are available at github.com/apapoudakis/qa-guided-reasoning.

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

🎓University of EdinburghEdinburgh · United Kingdom359 works · 77,352 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 2Research subjects and languages the work is about.Reasoning · 1,426 worksQuestion answering · 441 works

Cite

Cite as Argyrios Papoudakis, Mirella Lapata and Frank Keller. 2026. Think Before you Write: QA-Guided Reasoning for Character Descriptions in Books. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1259

@inproceedings{papoudakis2026think,
  title = {{Think Before you Write: QA-Guided Reasoning for Character Descriptions in Books}},
  author = {Argyrios Papoudakis, Mirella Lapata and Frank Keller},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1259},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1259}
}

References

66 references, 33 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. David Bamman, Olivia Lewke, and Anya Mansoor. 2020. An annotated dataset of coreference in English literature. In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 44–54, Marseille, France. European Language Resources Association.
  2. David Bamman, Brendan O’Connor, and Noah A. Smith. 2013. Learning latent personas of film characters. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 352–361, Sofia, Bulgaria. Association for Computational Linguistics.
  3. Sabyasachee Baruah and Shrikanth Narayanan. 2023. Character coreference resolution in movie screenplays. In Findings of the Association for Computational Linguistics: ACL 2023, pages 10300–10313, Toronto, Canada. Association for Computational Linguistics.
  4. Sabyasachee Baruah and Shrikanth Narayanan. 2025. CHATTER: A character-attribution dataset for narrative understanding. In Proceedings of the The 7th Workshop on Narrative Understanding, pages 52–63, Albuquerque, New Mexico. Association for Computational Linguistics. find
  5. Yao Fu. 2024. Challenges in deploying long-context transformers: A theoretical peak performance analysis. arXiv preprint arXiv:2405.08944. find
  6. Iz Beltagy, Matthew E Peters, and Arman Cohan. 2020. Longformer: The long-document transformer. arXiv preprint arXiv:2004.05150.
  7. Tianyu Gao, Alexander Wettig, Howard Yen, and Danqi Chen. 2025. How to train long-context language models (effectively). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7376–7399, Vienna, Austria. Association for Computational…
  8. Faeze Brahman, Meng Huang, Oyvind Tafjord, Chao Zhao, Mrinmaya Sachan, and Snigdha Chaturvedi. 2021. “let your characters tell their story”: A dataset for character-centric narrative understanding. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 1734–1752, Punta Cana,… find
  9. Yapei Chang, Yekyung Kim, Michael Krumdick, Amir Zadeh, Chuan Li, Chris Tanner, and Mohit Iyyer. 2025. Bleuberi: Bleu is a surprisingly effective reward for instruction following. arXiv preprint arXiv:2505.11080. find
  10. Yapei Chang, Kyle Lo, Tanya Goyal, and Mohit Iyyer. 2023. Booookscore: A systematic exploration of book-length summarization in the era of llms. In International Conference on Learning Representations. find
  11. Snigdha Chaturvedi, Shashank Srivastava, Hal Daume III, and Chris Dyer. 2015. Modeling dynamic relationships between characters in literary novels. arXiv preprint arXiv:1511.09376. — [arXiv:1511.09376](https://arxiv.org/abs/1511.09376) find
  12. Nuo Chen, Yan Wang, Haiyun Jiang, Deng Cai, Yuhan Li, Ziyang Chen, Longyue Wang, and Jia Li. 2023. Large language models meet harry potter: A dataset for aligning dialogue agents with characters. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 8506–8520, Singapore. As…
  13. Tri Dao, Dan Fu, Stefano Ermon, Atri Rudra, and Christopher Ré. 2022. Flashattention: Fast and memory-efficient exact attention with io-awareness. Advances in neural information processing systems, 35:16344–16359. find
  14. Daniel Deutsch, Tania Bedrax-Weiss, and Dan Roth. 2021. Towards question-answering as an automatic metric for evaluating the content quality of a summary. Transactions of the Association for Computational Linguistics, 9:774–789.
  15. Yiran Ding, Li Lyna Zhang, Chengruidong Zhang, Yuanyuan Xu, Ning Shang, Jiahang Xu, Fan Yang, and Mao Yang. 2024. Longrope: Extending llm context window beyond 2 million tokens. Preprint, arXiv:2402.13753.
  16. Angela Fan, Mike Lewis, and Yann Dauphin. 2018. Hierarchical neural story generation. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 889–898, Melbourne, Australia. Association for Computational Linguistics.
  17. Alexander Gurung and Mirella Lapata. 2024. CHIRON: Rich character representations in long-form narratives. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 8523–8547, Miami, Florida, USA. Association for Computational Linguistics. find
  18. Alexander Gurung and Mirella Lapata. 2025. Learning to reason for long-form story generation. arXiv preprint arXiv:2503.22828. — [arXiv:2503.22828](https://arxiv.org/abs/2503.22828) find
  19. Daniel Han, Michael Han, and Unsloth Team. 2023. Unsloth. find
  20. Pengcheng He, Jianfeng Gao, and Weizhu Chen. 2023. DeBERTav3: Improving deBERTa using ELECTRA-style pre-training with gradient-disentangled embedding sharing. In The Eleventh International Conference on Learning Representations.
  21. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, and 1 others. 2022. Lora: Low-rank adaptation of large language models. ICLR, 1(2):3.
  22. Jian Hu, Xibin Wu, Zilin Zhu, Weixun Wang, Dehao Zhang, Yu Cao, and 1 others. 2024. Openrlhf: An easy-to-use, scalable and high-performance rlhf framework. arXiv preprint arXiv:2405.11143.
  23. Fantine Huot, Joshua Maynez, Shashi Narayan, Reinald Kim Amplayo, Kuzman Ganchev, Annie Priyadarshini Louis, Anders Sandholm, Dipanjan Das, and Mirella Lapata. 2023. Text-blueprint: An interactive platform for plan-based conditional generation. In Proceedings of the 17th Conference of the European C… find
  24. Naoya Inoue, Charuta Pethe, Allen Kim, and Steven Skiena. 2022. Learning and evaluating character representations in novels. In Findings of the Association for Computational Linguistics: ACL 2022, pages 1008–1019, Dublin, Ireland. Association for Computational Linguistics. find
  25. Mohit Iyyer, Anupam Guha, Snigdha Chaturvedi, Jordan Boyd-Graber, and Hal Daumé III. 2016. Feuding families and former Friends: Unsupervised learning for dynamic fictional relationships. In Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguist… find
  26. Evgeny Kim and Roman Klinger. 2019. Frowning Frodo, wincing Leia, and a seriously great friendship: Learning to classify emotional relationships of fictional characters. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua… find
  27. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th symposium on operating systems principles, pages 611–626.
  28. Vincent Labatut and Xavier Bost. 2019. Extraction and analysis of fictional character networks: A survey. ACM Comput. Surv., 52(5). find
  29. Kunze Li and Yu Zhang. 2024. Planning first, question second: An LLM-guided method for controllable question generation. In Findings of the Association for Computational Linguistics: ACL 2024, pages 4715–4729, Bangkok, Thailand. Association for Computational Linguistics. find
  30. Tianle Li, Ge Zhang, Quy Duc Do, Xiang Yue, and Wenhu Chen. 2024. Long-context llms struggle with long in-context learning. arXiv preprint arXiv:2404.02060.
  31. Chin-Yew Lin. 2004. ROUGE: A package for automatic evaluation of summaries. In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics.
  32. Danyang Liu, Juntao Li, Meng-Hsuan Yu, Ziming Huang, Gongshen Liu, Dongyan Zhao, and Rui Yan. 2020. A character-centric neural model for automated story generation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 1725–1732. find
  33. Dongqi Liu, Xi Yu, Vera Demberg, and Mirella Lapata. 2025. Explanatory summarization with discourse-driven planning. Transactions of the Association for Computational Linguistics, 13:1146–1170.
  34. Nelson F. Liu, Kevin Lin, John Hewitt, Ashwin Paranjape, Michele Bevilacqua, Fabio Petroni, and Percy Liang. 2024. Lost in the middle: How language models use long contexts. Transactions of the Association for Computational Linguistics, 12:157–173.
  35. Louis Mahon and Mirella Lapata. 2024. A modular approach for multimodal summarization of TV shows. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8272–8291, Bangkok, Thailand. Association for Computational Linguistics. find
  36. Giuliano Martinelli, Tommaso Bonomo, Pere-Lluís Huguet Cabot, and Roberto Navigli. 2025. BOOK-COREF: Coreference resolution at book scale. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 24526–24544, Vienna, Austria. Associati… find
  37. Shashi Narayan, Joshua Maynez, Reinald Kim Amplayo, Kuzman Ganchev, Annie Louis, Fantine Huot, Anders Sandholm, Dipanjan Das, and Mirella Lapata. 2023. Conditional generation with a question-answering blueprint. Transactions of the Association for Computational Linguistics, 11:974–996.
  38. Kimia Noorbakhsh, Joseph Chandler, Pantea Karimi, Mohammad Alizadeh, and Hari Balakrishnan. 2025. Savaal: Scalable concept-driven question generation to enhance human learning. arXiv preprint arXiv:2502.12477. — [arXiv:2502.12477](https://arxiv.org/abs/2502.12477) find
  39. Eric W Noreen. 1989. Computer intensive methods for hypothesis testing: An introduction. Wiley, New York, 19:21. find
  40. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, and 1 others. 2022. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35:27730–27744.
  41. Argyrios Papoudakis, Mirella Lapata, and Frank Keller. 2024. BookWorm: A dataset for character description and analysis. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 4471–4500, Miami, Florida, USA. Association for Computational Linguistics. find
  42. Pranav Rajpurkar, Robin Jia, and Percy Liang. 2018. Know what you don’t know: Unanswerable questions for SQuAD. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 784–789, Melbourne, Australia. Association for Computational Ling…
  43. David Rein, Betty Li Hou, Asa Cooper Stickland, Jackson Petty, Richard Yuanzhe Pang, Julien Dirani, Julian Michael, and Samuel R Bowman. 2024. Gpqa: A graduate-level google-proof q&a benchmark. In First Conference on Language Modeling.
  44. Yisi Sang, Xiangyang Mou, Mo Yu, Dakuo Wang, Jing Li, and Jeffrey Stanton. 2022. MBTI personality prediction for fictional characters using movie scripts. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 6715–6724, Abu Dhabi, United Arab Emirates. Association for Compu… find
  45. Yunfan Shao, Linyang Li, Junqi Dai, and Xipeng Qiu. 2023. Character-LLM: A trainable agent for role-playing. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 13153–13187, Singapore. Association for Computational Linguistics.
  46. Krishnapriya Vishnubhotla, Adam Hammond, Graeme Hirst, and Saif Mohammad. 2024. The emotion dynamics of literary novels. In Findings of the Association for Computational Linguistics: ACL 2024, pages 2557–2574, Bangkok, Thailand. Association for Computational Linguistics. find
  47. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, and 1 others. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300.
  48. Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. 2021. Finetuned language models are zero-shot learners. arXiv preprint arXiv:2109.01652.
  49. Parshin Shojaee, Aneesh Jain, Sindhu Tipirneni, and Chandan K Reddy. 2023. Execution-based code generation using deep reinforcement learning. arXiv preprint arXiv:2301.13816. find
  50. 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. Advances in neural information processing systems, 35:24824– 24837.
  51. Marcin Skowron, Martin Trapp, Sabine Payr, and Robert Trappl. 2016. Automatic identification of character types from film dialogs. Applied Artificial Intelligence, 30(10):942–973. find
  52. Jeff Wu, Long Ouyang, Daniel M Ziegler, Nisan Stiennon, Ryan Lowe, Jan Leike, and Paul Christiano. 2021. Recursively summarizing books with human feedback. arXiv preprint arXiv:2109.10862.
  53. Yixiao Song, Yekyung Kim, and Mohit Iyyer. 2024. VeriScore: Evaluating the factuality of verifiable claims in long-form text generation. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 9447–9474, Miami, Florida, USA. Association for Computational Linguistics.
  54. Peng Xu, Wei Ping, Xianchao Wu, Lawrence McAfee, Chen Zhu, Zihan Liu, Sandeep Subramanian, Evelina Bakhturina, Mohammad Shoeybi, and Bryan Catanzaro. 2023. Retrieval meets long context large language models. arXiv preprint arXiv:2310.03025.
  55. Liyan Tang, Philippe Laban, and Greg Durrett. 2024. MiniCheck: Efficient fact-checking of LLMs on grounding documents. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 8818–8847, Miami, Florida, USA. Association for Computational Linguistics.
  56. Gemini Team, Petko Georgiev, Ving Ian Lei, Ryan Burnell, Libin Bai, Anmol Gulati, Garrett Tanzer, Damien Vincent, Zhufeng Pan, Shibo Wang, and 1 others. 2024. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context. arXiv preprint arXiv:2403.05530.
  57. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  58. An Yang, Beichen Zhang, Binyuan Hui, Bofei Gao, Bowen Yu, Chengpeng Li, Dayiheng Liu, Jianhong Tu, Jingren Zhou, Junyang Lin, and 1 others. 2024. Qwen2. 5-math technical report: Toward mathematical expert model via self-improvement. arXiv preprint arXiv:2409.12122. — Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvement
  59. Gemma Team, Aishwarya Kamath, Johan Ferret, Shreya Pathak, Nino Vieillard, Ramona Merhej, Sarah Perrin, Tatiana Matejovicova, Alexandre Ramé, Morgane Rivière, and 1 others. 2025. Gemma 3 technical report. arXiv preprint arXiv:2503.19786. find
  60. Mo Yu, Jiangnan Li, Shunyu Yao, Wenjie Pang, Xiaochen Zhou, Zhou Xiao, Fandong Meng, and Jie Zhou. 2023. Personality understanding of fictional characters during book reading. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 14… find
  61. Runchu Tian, Yanghao Li, Yuepeng Fu, Siyang Deng, Qinyu Luo, Cheng Qian, Shuo Wang, Xin Cong, Zhong Zhang, Yesai Wu, Yankai Lin, Huadong Wang, and Xiaojiang Liu. 2025. Distance between relevant information pieces causes bias in long-context LLMs. In Findings of the Association for Computational Ling… find
  62. Xinfeng Yuan, Siyu Yuan, Yuhan Cui, Tianhe Lin, Xintao Wang, Rui Xu, Jiangjie Chen, and Deqing Yang. 2024. Evaluating character understanding of large language models via character profiling from fictional works. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processi… find
  63. Jonathan Uesato, Nate Kushman, Ramana Kumar, Francis Song, Noah Siegel, Lisa Wang, Antonia Creswell, Geoffrey Irving, and Irina Higgins. 2022. Solving math word problems with process-and outcome-based feedback. arXiv preprint arXiv:2211.14275.
  64. Weiwei Zhang, Jackie Chi Kit Cheung, and Joel Oren. 2019. Generating character descriptions for automatic summarization of fiction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 7476–7483. find
  65. Xuandong Zhao, Zhewei Kang, Aosong Feng, Sergey Levine, and Dawn Song. 2025. Learning to reason without external rewards. arXiv preprint arXiv:2505.19590. find
  66. Lixing Zhu, Runcong Zhao, Lin Gui, and Yulan He. 2023. Are NLP models good at tracing thoughts: An overview of narrative understanding. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 10098–10121, Singapore. Association for Computational Linguistics. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/papoudakis-2026-think-before-you-write-qa/