Authors : Soham Sachin Shelar; Dr. Manisha Bharati

Volume/Issue : Volume 11 - 2026, Issue 5 - May

---

Google Scholar : https://tinyurl.com/55f6drj9

Scribd : https://tinyurl.com/yfrhvcss

DOI : https://doi.org/10.38124/ijisrt/26May1714

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

---

Abstract : The Indian legal system generates an enormous volume of judgments every year across thousands of courts, yet access to structured legal research remains limited for a large portion of the population. Existing large language models (LLMs), while capable of impressive natural language generation, suffer from hallucination—fabricating statutory provisions, inventing case citations, and producing reasoning that lacks grounding in actual evidence. These failures are especially dangerous in the legal domain, where an incorrect citation can invalidate an entire argument. This paper presents LegalMind AI, a multi-agent legal reasoning system designed specifically for Indian law.

Keywords : Multi-Agent AI; Legal Reasoning; Retrieval-Augmented Generation; QLoRA Fine-Tuning; Indian Law; Natural Language Inference; Hallucination Detection; Mistral-7B; FAISS; IL-TUR Dataset; DeBERTa; Streamlit.

References :

1. N. Aletras et al., “Predicting judicial decisions of the European Court of Human Rights,” PeerJ Computer Science, vol. 2, p. e93, 2016.
2. K. Malik et al., “ILDC for CJPE,” in Proc. ACL, 2021, pp. 4046–4062.
3. A. Jain et al., “Predicting and Explaining Indian Court Decisions,” in Proc. EMNLP Findings, 2021.
4. Exploration-Lab, “IL-TUR Benchmark,” Hugging-
5. Face Datasets: Exploration-Lab/IL-TUR, 2022.
6. J. Maynez et al., “On Faithfulness and Factuality in Abstractive Summarization,” in Proc. ACL, 2020, pp. 1906–1919.
7. Z. Ji et al., “Survey of Hallucination in NLG,” ACM Computing Surveys, vol. 55, no. 12, pp. 1–38, 2023.
8. Y. Bang et al., “A Multitask Evaluation of ChatGPT,” arXiv:2302.04023, 2023.
9. P. Lewis et al., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks,” in Advances in NeurIPS, 2020.
10. V. Karpukhin et al., “Dense Passage Retrieval,” in Proc. EMNLP, 2020.
11. N. Reimers and I. Gurevych, “Sentence-BERT,” in Proc. EMNLP, 2019.
12. R. Nogueira and K. Cho, “Passage Re-ranking with BERT,” arXiv:1901.04085, 2019.
13. J. Johnson, M. Douze, and H. Jégou, “Billion-Scale Similarity Search with GPUs,” IEEE Trans. Big Data, vol. 7, no. 3, pp. 535–547, 2021.
14. I. Chalkidis et al., “LEGAL-BERT,” in Findings of EMNLP, 2020.
15. J. Wei et al., “Finetuned Language Models Are ZeroShot Learners,” in Proc. ICLR, 2022.
16. E. Hu et al., “LoRA,” in Proc. ICLR, 2022.
17. T. Dettmers et al., “QLoRA,” in Advances in NeurIPS, 2023.
18. A. Jiang et al., “Mistral 7B,” arXiv:2310.06825, 2023. [18] P. He et al., “DeBERTa,” in Proc. ICLR, 2021.
19. S. Bowman et al., “A Large Annotated Corpus for NLI,” in Proc. EMNLP, 2015, pp. 632–642.
20. A. Williams, N. Nangia, and S. Bowman, “A BroadCoverage Challenge Corpus for Sentence Understanding,” in Proc. NAACL, 2018.
21. J. Park et al., “Generative Agents,” in Proc. UIST, 2023.
22. Q. Wu et al., “AutoGen,” arXiv:2308.08155, 2023.
23. C. Chan et al., “ChatEval,” arXiv:2308.07201, 2023.
24. A. Vaswani et al., “Attention Is All You Need,” in Advances in NeurIPS, 2017.
25. J. Devlin et al., “BERT,” in Proc. NAACL, 2019.
26. H. Touvron et al., “Llama 2,” arXiv:2307.09288, 2023.
27. Z. Guo et al., “A Survey on Automated FactChecking,” TACL, vol. 10, pp. 178–206, 2022.
28. J. Zhou et al., “LawGPT,” arXiv:2306.03061, 2023.
29. S. Yue et al., “Disc-LawLLM,” arXiv:2309.11325, 2023.
30. T. Wolf et al., “Transformers,” in Proc. EMNLP System Demonstrations, 2020, pp. 38–45.