Civil Court Simulation with Large Language Models

2026-06-08

AI Summary

This study addresses the limitations of existing courtroom simulations, which predominantly focus on criminal cases and struggle to capture the flexible claim structures, liability attribution, and adjudicative reasoning characteristic of civil litigation, while also suffering from high manual costs and poor scalability. To overcome these challenges, this work proposes the first large language model–based multi-agent framework for simulating civil court proceedings in accordance with the five-stage structure of Chinese civil trials. The framework incorporates a memory module and a legal provision retrieval mechanism to support long-horizon legal reasoning across the entire litigation process. It further introduces an innovative five-dimensional reliability assessment model—encompassing legal basis, information conditions, judicial competence, role positioning, and socio-legal context. Experimental results demonstrate strong performance in liability allocation and multi-claim adjudication, with memory quality significantly influencing simulation fidelity, thereby validating the framework’s effectiveness and feasibility.

Abstract

Court simulation bridges legal education and judicial practice, yet human-based simulations are costly and difficult to scale. Large language models (LLMs) offer a scalable alternative, but existing court-simulation research mainly focuses on criminal cases. Civil litigation is more common in practice and harder to simulate because its claims, liability, and remedies are more flexible. We present a multi-agent court simulation framework for Chinese civil cases. The framework organizes role-based interaction through a five-stage civil trial procedure and integrates memory module and statute retrieval to support long-process adjudication. Experiments show that the framework produces reliable civil judgments, with clear strengths in liability allocation and multi-item adjudication. Further experiments show that memory quality substantially affects downstream simulation quality. Through a five-layer factor framework, we analyze how legal grounding, information conditions, judicial capability and role orientation, organizational pressure, and social context affect the framework's reliability and behavior. These results support the effectiveness of the proposed framework for civil court simulation. The dataset and code are available at: https://github.com/foggpoy/Civil-Court.

Problem

• civil litigation
• court simulation
• large language models
• liability allocation
• multi-item adjudication

Innovation

• multi-agent simulation
• civil litigation
• large language models
• legal reasoning
• memory-augmented retrieval