From Query to Counsel: Structured Reasoning with a Multi-Agent Framework and Dataset for Legal Consultation

2026-04-12

AI Summary

This study addresses the challenges of legal consultation QA, including scarce high-quality data, task complexity, and strong contextual dependencies. The authors propose a structured task decomposition approach grounded in a legal element graph, modeling queries as graph structures that integrate entities, events, intents, and legal issues. They introduce JurisMA, a modular and interpretable multi-agent reasoning framework incorporating dynamic routing, statute grounding, and style optimization mechanisms. Trained on JurisCQAD—a large-scale, self-constructed Chinese legal consultation dataset—the system significantly outperforms both general-purpose and legal-domain large language models across multiple semantic and lexical metrics on the LawBench benchmark.

Technology Category

• Natural Language Processing: Question Answering
• Multiagent Systems: Modeling other Agents
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Legal consultation question answering (Legal CQA) presents unique challenges compared to traditional legal QA tasks, including the scarcity of high-quality training data, complex task composition, and strong contextual dependencies. To address these, we construct JurisCQAD, a large-scale dataset of over 43,000 real-world Chinese legal queries annotated with expert-validated positive and negative responses, and design a structured task decomposition that converts each query into a legal element graph integrating entities, events, intents, and legal issues. We further propose JurisMA, a modular multi-agent framework supporting dynamic routing, statutory grounding, and stylistic optimization. Combined with the element graph, the framework enables strong context-aware reasoning, effectively capturing dependencies across legal facts, norms, and procedural logic. Trained on JurisCQAD and evaluated on a refined LawBench, our system significantly outperforms both general-purpose and legal-domain LLMs across multiple lexical and semantic metrics, demonstrating the benefits of interpretable decomposition and modular collaboration in Legal CQA.

Problem

• Legal consultation question answering
• data scarcity
• task complexity
• contextual dependency

Innovation

• legal consultation
• multi-agent framework
• structured reasoning
• legal element graph
• dataset construction