International Journal of Innovative Research in Computer and Communication Engineering
ISSN Approved Journal | Impact factor: 8.771 | ESTD: 2013 | Follows UGC CARE Journal Norms and Guidelines
| Monthly, Peer-Reviewed, Refereed, Scholarly, Multidisciplinary and Open Access Journal | High Impact Factor 8.771 (Calculated by Google Scholar and Semantic Scholar | AI-Powered Research Tool | Indexing in all Major Database & Metadata, Citation Generator | Digital Object Identifier (DOI) |
| TITLE | Justice: A Hybrid Retrieval-Augmented Generation and Knowledge Graph Framework for Indian Legal Query Resolution |
|---|---|
| ABSTRACT | Access to legal information in India has been challenging because of the sheer volume of judgments issued by the Supreme Court of more than 70 years, the difficulty in comprehending legal language, and multilinguism. In our work, we propose Justice AI, a Legal Information Intelligent Assistant powered by a new architecture that blends retrieval-augmentation-generation and Neo4j Knowledge Graph. It analyzes 26,688 Supreme Court judgments from 1950 to 2025, employs a two-tier process to extract structured data using deterministic regex heuristics and Groq-optimized LLaMA-3.3-70B model, and indexes data across six vector databases of Pinecone and one graph database of Neo4j AuraDB. The query is processed through 9 paths of parallel retrieval encompassing semantic vector similarity searches and Cypher graph traversal. The hybrid approach delivers 82% precision for structured legal queries, hallucination rate reduction from 18% to 4%, and multilingual responses within 3.2 seconds in English, Hindi, and Marathi languages. Comparative studies with Indian Kanoon, JUDIS, ungrounded GPT-4, LEGAL-BERT, and InLegalBERT demonstrate clear advantages in every metric studied. |
| AUTHOR | ROHAN SALUNKHE, RAJWARDHAN JAGTAP, VAISHNAV CHOUGULE, KIRAN GORE Undergraduate Student, Dept. of Information Technology, G.H. Raisoni College of Engineering and Management, Pune, India Associate Professor, Dept. of Information Technology G.H. Raisoni College of Engineering and Management, Pune, India |
| VOLUME | 187 |
| DOI | DOI: 10.15680/IJIRCCE.2026.1408014 |
| pdf/14_Justice A Hybrid Retrieval-Augmented Generation and Knowledge Graph Framework for Indian Legal Query Resolution.pdf | |
| KEYWORDS | |
| References | [1] P. Lewis, E. Perez, A. Piktus, F. Petroni, V. Karpukhin, N. Goyal, H. Kuttler, M. Lewis, W.-T. Yih, T. Rockta¨schel, S. Riedel, and D. Kiela, “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), vol. 33, pp. 9459–9474, 2020. [2] D. Maxwell and J. Schafer, “Legal information retrieval: A survey of systems and approaches,” Legal Inf. Manage., vol. 13, no. 4, pp. 228–239, 2013. [3] I. Chalkidis, M. Fergadiotis, P. Malakasiotis, N. Aletras, and I. Androutsopoulos, “LEGAL-BERT: The Muppets straight out of Law School,” in Findings of EMNLP 2020, pp. 2898–2904, 2020. [4] S. Paul and P. Goyal, “Pre-training Transformers on Indian Legal Text,” in Proc. Natural Legal Lang. Process. Workshop (NLLP), pp. 61–70, 2022. [5] G. Izacard and E. Grave, “Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering,” in Proc. EACL, pp. 874–880, 2021. [6] N. Guha et al., “LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in LLMs,” in Proc. NeurIPS, 2023. [7] M. Koniaris, I. Anagnostopoulos, and Y. Vassiliou, “Network Analysis in the Legal Domain: A Complex Model for European Union Legal Sources,” J. Complex Netw., vol. 6, no. 2, pp. 243–268, 2017. [8] W. Lam, H. T. Poon, and C. T. Yu, “Knowledge Graph Enhanced Legal Judgment Prediction via Multi-Task Learning,” in Proc. CIKM, pp. 917–926, 2021. [9] J. Niklaus, I. Chalkidis, and M. Stu¨rmer, “Swiss-Judgment-Prediction: A Multilingual Legal Judgment Prediction Benchmark,” in Proc. NLLP Workshop, pp. 19–35, 2021. [10] A. Singh, “Legal Dataset: SC Judgments India 1950–2024,” Kaggle, 2024. [Online]. Available: https://www.kaggle.com/datasets/ adarshsingh0903/legal-dataset-sc-judgments-india-19502024 [11] H. Touvron et al., “Llama 2: Open Foundation and Fine-Tuned Chat Models,” arXiv preprint arXiv:2307.09288, 2023. [12] L. Wang, N. Yang, X. Huang, L. Yang, R. Majumder, and F. Wei, “Multilingual E5 Text Embeddings: A Technical Report,” arXiv preprint arXiv:2402.05672, 2024. [13] S. Robertson and H. Zaragoza, “The Probabilistic Relevance Framework: BM25 and Beyond,” Found. Trends Inf. Retr., vol. 3, no. 4, pp. 333–389, 2009. [14] G. Zheng, S. Guo, Y. Liu, and K. Q. Zhu, “Does BERT Understand Law? An Empirical Study of Domain Adaptation in Legal NLP,” in Proc. ICAIL, pp. 251–260, 2021. [15] W. Shi, S. Min, M. Yasunaga, M. Seo, R. James, M. Lewis, L. Zettlemoyer, and W.-T. Yih, “REPLUG: Retrieval-Augmented Black-Box Language Models,” arXiv preprint arXiv:2301.12652, 2023. [16] J. Savelka, K. D. Ashley, M. Gray, H. Westermann, and H. Xu, “Explaining Legal Concepts with Augmented Large Language Models (GPT-4),” in Proc. ICAIL, pp. 41–50, 2023. [17] F. de Melo Gomes, A. Wyner, and J. F. Baget, “Knowledge Graphs for Legislation: Understanding and Connecting Legal Texts,” in Proc. Joint Ontol. Workshops (JOWO), 2020. [18] A. Blair, J. Chen, and M. Katz, “GPT-4 Passes the Bar Exam,” arXiv preprint arXiv:2212.14402, 2023. [19] J. J. Nay, “Large Language Models as Tax Attorneys: A Case Study in Legal Capabilities Emergence,” Philos. Trans. R. Soc. A, vol. 381, no. 2251, 2023. [20] National Informatics Centre, “JUDIS: Judgment Information System,” Supreme Court of India, 2024. [Online]. Available: https://judis.nic.in [21] Groq Inc., “Groq LPU Inference Engine,” 2024. [Online]. Available: https://groq.com |
