The complexity of Vietnamese legal documents poses significant challenges in accessing and comprehending legal information. This paper introduces VNLegalEase, an advanced chatbot designed to address these challenges in the Vietnamese legal context. Leveraging state-of-the-art natural language processing (NLP) techniques, VNLegalEase integrates GPT-4o, LlamaIndex, vector databases, and Retrieval-Augmented Generation (RAG) to deliver accurate and context-specific responses to legal queries. Key contributions include: (1) development of a Vietnamese legal-specific chatbot; (2) novel integration of advanced technologies for enhanced legal information retrieval and response generation; (3) creation of a comprehensive Vietnamese legal dataset comprising 36,602 legal documents and 300,000 Q&A pairs; and (4) rigorous evaluation demonstrating a 12% increase in accuracy and 52% reduction in response time compared to baseline models. Our experimental results show VNLegalEase achieves 87% accuracy on complex legal queries and 90% coverage of Vietnamese legal knowledge across 11 categories. User studies indicate high satisfaction, with 60% of users rating response accuracy at 4 or 5 on a 5-point Likert scale. This research advances legal AI systems for the Vietnamese context and provides a framework for developing similar systems in other languages and jurisdictions.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

VNLegalEase: A Vietnamese Legal Query Chatbot

  • Pham Thi Xuan Hien,
  • Nguyen Thanh Tuong Vy,
  • Huu-Dung Ngo

摘要

The complexity of Vietnamese legal documents poses significant challenges in accessing and comprehending legal information. This paper introduces VNLegalEase, an advanced chatbot designed to address these challenges in the Vietnamese legal context. Leveraging state-of-the-art natural language processing (NLP) techniques, VNLegalEase integrates GPT-4o, LlamaIndex, vector databases, and Retrieval-Augmented Generation (RAG) to deliver accurate and context-specific responses to legal queries. Key contributions include: (1) development of a Vietnamese legal-specific chatbot; (2) novel integration of advanced technologies for enhanced legal information retrieval and response generation; (3) creation of a comprehensive Vietnamese legal dataset comprising 36,602 legal documents and 300,000 Q&A pairs; and (4) rigorous evaluation demonstrating a 12% increase in accuracy and 52% reduction in response time compared to baseline models. Our experimental results show VNLegalEase achieves 87% accuracy on complex legal queries and 90% coverage of Vietnamese legal knowledge across 11 categories. User studies indicate high satisfaction, with 60% of users rating response accuracy at 4 or 5 on a 5-point Likert scale. This research advances legal AI systems for the Vietnamese context and provides a framework for developing similar systems in other languages and jurisdictions.