The week's pick
Random Articles
Reseach Article
Evaluation of VidhiAI an AI- powered System for Legal Document Summarization
| International Journal of Computer Applications |
| Foundation of Computer Science (FCS), NY, USA |
| Volume 187 - Number 117 |
| Year of Publication: 2026 |
| Authors: Leishemba Okram, Snehasish Bose, Babu S. |
10.5120/ijca1ec1b50a253f
|
Leishemba Okram, Snehasish Bose, Babu S. . Evaluation of VidhiAI an AI- powered System for Legal Document Summarization. International Journal of Computer Applications. 187, 117 ( Jun 2026), 19-27. DOI=10.5120/ijca1ec1b50a253f
Abstract
By increasing the effectiveness of legal research, documentation and compliance verification, artificial intelligence is revolutionizing the legal industry. In this work, Vidhi AI is provided as a domain-specific large language model (LLM) that can accurately and contextually summarize legal documents and perform statutory and compliance checks. Vidhi AI is optimized on carefully selected legal datasets, in contrast to general purpose LLMs, guaranteeing accuracy and pertinence in legal applications. The model creates structured summaries of legal documents and cross-checks statutory compliance against jurisdictional requirements using a hybrid technique that combines refined LLMs with Retrieval Augmented Generation (RAG). Along with a comparison to other legal AI tools now in use, the study outlined the model architecture, data curation procedure, and evaluation methodology. The difficulties in developing legal AI are also covered in the study, such as data bases, ethical issues, and regulatory limitations. Vidhi AI is a step toward intelligent legal assistants that support attorneys while guaranteeing compliance with the law, improve document analysis, and increase compliance verification.
References
- Chouhan, A. and Gertz, M.(2024). LexDrafter: “Terminology Drafting for Legislative Documents using Retrieval Augmented Generation”, arXiv:2403.16295v1 [cs.CL].
- Lewis P. et al. (2020). “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”, inProc.34thInt. Conf. NeuralInf. Process. Syst.(NeurIPS’20), 793, 9459–9474, ACM.
- Zhang, T. et al. (2020). “BERTScore: Evaluating Text Generation with BERT”, in Proc.8thInt.Conf.Learn.Representations (ICLR’20), Open Review.net.
- “EUR-Lex platform,” accessed on Sep. 11, 2023. [Online]. Available:https://eur-lex.europa.eu
- “Indian Legal Documents Corpus (ILDC),” [Online]. Available: https://github.com/Legal-NLP-EkStep/ILDC
- European Commission and Legal Service. (2015). Joint Practical Guide of theEuropean Parliament, the Council and the Commission for PersonsInvolved in the Drafting of European Union Legislation. Publications Office.
- Sai, C. et al. (2023). “Identification and Visualization of Legal Definitions andLegal TermRelations,” in Adv. Concept. Model., 151-161.
- Waltl, B. et al. (2017). “Automated extraction of semantic information fromGermanlegal documents,” inIRIS: Int. Rechts inf. Symp., Austria.
- Huang, L. et al. (2023). “A survey on hallucination in large language models:Principles, taxonomy, challenges, and open questions,” Preprint.
- Hicks, M., Humphries, J. and Slate,r J. (2024). “ChatGPT is bullshit,” Ethics Inf. Technol., 26, 1-10.
- Dahl, M. et al. (2024). “Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models,” J. Legal Anal., 16, 64-93.
- Stu, S., Cyphert A. and Perl S. J. (2024). “Artificial Intelligence: Legal Reasoning, Legal Research and Legal Writing,” Minnesota J. Law, Sci. Technol., 25, 105-124.
- Ding, N. et al. (2023). “Parameter-efficient fine-tuning of large-scale pre-trained language models”. Nat. Mach. Intell.,5,220-235.
Index Terms
Keywords