Multilingual Ai-based Legal Document Analyzer Using Retrieval-augmented Generation And Transformer Models

Abstract

The Interpretation Of Legal Documents Remains A Complex, Time-intensive Challenge For Both Legal Professionals And The General Public. This Paper Presents A Multilingual AI-Based Legal Document Analyzer That Lever- Ages Retrieval-Augmented Generation (RAG), Transformer- Based Natural Language Processing (NLP), And Multilingual Translation Models To Automate The Analysis Of Legal Con- Tracts And Agreements. The Proposed System Integrates A Clause Extraction Engine Built On Legal-BERT, A Semantic Question-answering Module Powered By FAISS-indexed Vector Retrieval And Flan-T5 Generation, A BART-based Document Summarizer, And A Multilingual Translation Pipeline Supporting English, Hindi, Tamil, And Telugu. Deployed Through An Interactive Streamlit Web Interface, The Platform Enables Users To Upload PDF Documents And Receive Real- Time Clause Highlights, Contextual Answers, Concise Sum- Maries, And Cross-lingual Translations. Experimental Evaluation On A Diverse Corpus Of Legal Documents Demonstrates Clause Extraction Precision Of 92%, Question-answering Ac- Curacy Of 88%, And Sub-1.5-second Response Latency, With 93% Of Survey Respondents Rating The Interface As Intuitive. The System’s Modular Architecture Supports Continuous Improvement Via Active Learning From User Feedback And Plug- And-play Model Upgrades.

Keywords