Elamur Research⌕
☰
home / papers / 2026 / ArgAssist: LLM-based Argument Synthesis for Insurance Disputes

ArgAssist: LLM-based Argument Synthesis for Insurance Disputes

Summary

The paper addresses a gap in alternative dispute resolution (ADR): parties in insurance conflicts often cannot afford lawyers, yet building legally sound arguments normally requires professional expertise. The authors present ArgAssist, a module of a dispute resolution platform under development at TCS Research. It generates new arguments for either side, the insurer or the insured, instead of extracting them from existing documents. Each argument is a claim backed by typed premises: fact, precedent, statute, evidence/testimony or world knowledge.

The pipeline has two stages. First, an LLM drafts base arguments from structured case information: agreed facts, disputed aspects, evidence, testimonies and demands. Second, these drafts are strengthened with retrieved prior cases and with statutes, which the authors define broadly to include insurance-policy clauses.

The authors also propose ArgMatch, a metric that scores generated arguments against expert-written ones. On two Indian consumer-court insurance datasets, ArgAssist beats a fine-tuning baseline and its own ungrounded drafts in every evaluation setting. The results suggest that explicit discourse structure plus legal grounding produces arguments closer to real court submissions.

Key points

  • Data and model
    • Two datasets, VID and LID, were built from insurance disputes scraped from the NCDRC website.
    • Precedents come from a pool (PC) of 211 Supreme Court of India judgements.
    • All experiments use GPT-4.1 only.
  • Prior-case retrieval
    • A prior case is ranked by a weighted sum of embedding cosine similarities with the input case.
    • The similarities are computed separately for facts, disagreement aspects, evidences and testimonies.
  • ArgMatch metric
    • Claims of gold and generated arguments are paired by bipartite matching.
    • Each matched pair is scored by combining claim and premise similarity with weight λ (default 0.5) and a threshold of 0.6.
    • The metric reports precision, recall and F1, and can optionally require premise types to match.
  • Results
    • The final ArgAssist output recovers about 53–65% of the arguments actually presented in court (ArgMatch recall).
    • Precision is lower because the system produces extra arguments with no gold counterpart.
    • Qualitative scoring rated these extra arguments as reasonably good.
  • Baseline and future work
    • The baseline follows Tuvey and Sen (2023), who fine-tuned GPT-2 and Flan-T5.
    • Planned work covers other domains (e-commerce, taxation), legal factors as an additional premise type, and more LLMs.

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

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.sigdial-1.9

@inproceedings{sinha2026argassist,
  title = {{ArgAssist: LLM-based Argument Synthesis for Insurance Disputes}},
  author = {Sinha, Anubhav and Ramrakhiyani, Nitin and Pawar, Sachin and Narang, Isha and Apte, Manoj},
  booktitle = {Proceedings of the 27th Annual Meeting of the Special Interest Group on Discourse and Dialogue},
  year = {2026},
  doi = {10.18653/v1/2026.sigdial-1.9},
  url = {https://doi.org/10.18653/v1/2026.sigdial-1.9}
}

References

25 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sergio Torres Aguilar. 2025. "don’t teach minerva": Guiding llms through complex syntax for faithful latin translation with rag. arXiv preprint arXiv:2511.01454. find
  2. Vincent Aleven. 2003. Using background knowledge in case-based legal reasoning: A computational model and an intelligent learning environment. Artificial Intelligence, 150(1):183–237. AI and Law. find
  3. Basit Ali, Sachin Pawar, Girish Palshikar, and Rituraj Singh. 2022. Constructing a dataset of support and attack relations in legal arguments in court judgements using linguistic rules. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 491–500, Marseille, France. E… find
  4. Basit Ali, Sachin Pawar, Girish Palshikar, Anindita Sinha Banerjee, and Dhirendra Singh. 2023. Legal argument extraction from court judgements using integer linear programming. In Proceedings of the 10th Workshop on Argument Mining, pages 52–63, Singapore. Association for Computational Linguistics. find
  5. Basit Ali, Anubhav Sinha, Nitin Ramrakhiyani, Sachin Pawar, Girish Keshav Palshikar, and Manoj Apte. 2026. Argumentation and judgement factors: LLM-based discovery and application in insurance disputes. In Proceedings of the 19th Conference of the European Chapter of the Association for Computationa… find
  6. David Ashton. 2024. Alternative dispute resolution. In European Parliamentary Research Service (EPRS). find
  7. Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Albert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Alex Castro-Ros, Marie Pellat, Kevin Robinson, and 16 others. 2022.… find
  8. Xinyu Hua, Zhe Hu, and Lu Wang. 2019. Argument generation with retrieval, planning, and realization. arXiv preprint arXiv:1906.03717. find
  9. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, and 1 others. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural information processing systems, 33:9459–… find
  10. Hao Li, Viktor Schlegel, Yizheng Sun, Riza Batista-Navarro, and Goran Nenadic. 2025. Large language models in argument mining: A survey. arXiv preprint arXiv:2506.16383. — [arXiv:2506.16383](https://arxiv.org/abs/2506.16383) find
  11. Kaveh Eskandari Miandoab and Vasanth Sarathy. 2024. “let’s argue both sides”: Argument generation can force small models to utilize previously inaccessible reasoning capabilities. In Proceedings of the 1st Workshop on Customizable NLP: Progress and Challenges in Customizing NLP for a Domain, Applica… find
  12. NITI-Aayog. 2021. Designing the Future of Dispute Resolution: The ODR Policy Plan for India. In The NITI Aayog Expert Committee Report on ODR. find
  13. Raquel Mochales Palau and Marie-Francine Moens. 2009. Argumentation mining: the detection, classification and structure of arguments in text. In Proceedings of the 12th international conference on artificial intelligence and law, pages 98–107. find
  14. Sachin Pawar, Manoj Apte, Girish Keshav Palshikar, Basit Ali, and Nitin Ramrakhiyani. 2025. DRAssist: Dispute Resolution Assistance using Large Language Models. In Proceedings of the Twentieth International Conference on Artificial Intelligence and Law (ICAIL), pages 188–198. find
  15. Mohamed Elaraby and Diane Litman. 2022. Arglegal-summ: Improving abstractive summarization of legal documents with argument mining. arXiv preprint arXiv:2209.01650. — [arXiv:2209.01650](https://arxiv.org/abs/2209.01650) find
  16. Lucile Favero, Daniel Frases, Juan Antonio Pérez-Ortiz, and Tanja Käser. 2025. ELLIS Alicante at CQs-gen 2025: Winning the critical thinking questions shared task: LLM-based question generation and selection. In Proceedings of the 12th Argument mining Workshop, pages 322–331, Vienna, Austria. Associ… find
  17. Morgan Gray, Jaromir Savelka, Wesley Oliver, and Kevin Ashley. 2024. Using llms to discover legal factors. arXiv preprint arXiv:2410.07504. find
  18. Morgan Gray, Li Zhang, and Kevin D Ashley. 2025. Generating case-based legal arguments with llms. In Proceedings of the 2025 Symposium on Computer Science and Law, pages 160–168. find
  19. Alexander Rogiers, Sander Noels, Maarten Buyl, and Tijl De Bie. 2024. Persuasion with large language models: a survey. Preprint, arXiv:2411.06837. — [arXiv:2411.06837](https://arxiv.org/abs/2411.06837) find
  20. Sougata Saha and Rohini K Srihari. 2023. ArgU: A controllable factual argument generator. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8373–8388. find
  21. Oscar Tuvey and Procheta Sen. 2023. Automated argument generation from legal facts. Preprint, arXiv:2310.05680. find
  22. Preetika Verma, Kokil Jaidka, and Svetlana Churina. 2025. “reasoning” with rhetoric: On the style-evidence tradeoff in llm-generated counter-arguments. In Proceedings of the International AAAI Conference on Web and Social Media, volume 19, pages 1966–1989. find
  23. Douglas Walton, Christopher Reed, and Fabrizio Macagno. 2008. Argumentation schemes. Cambridge University Press. find
  24. Gechuan Zhang, Paul Nulty, and David Lillis. 2023. Argument mining with graph representation learning. In Proceedings of the Nineteenth International Conference on Artificial Intelligence and Law, pages 371–380. find
  25. Li Zhang and Kevin D Ashley. 2025. Mitigating manipulation and enhancing persuasion: A reflective multi-agent approach for legal argument generation. arXiv preprint arXiv:2506.02992. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sinha-2026-argassist-llm-based-argument-synthesis/