Logic-Guided Multistage Inference for Explainable Multidefendant Judgment Prediction

2026-01-19

AI Summary

This study addresses the challenge posed by ambiguous judicial language in multi-defendant cases, which obscures role distinctions and compromises both the fairness of culpability attribution and the accuracy of AI-driven legal analysis. To resolve this, we propose the Masked Multi-Stage Inference (MMSI) framework, which explicitly integrates sentencing logic into a pre-trained Transformer encoder for the first time. MMSI employs a directional masking mechanism to clarify defendant roles, enhances sensitivity to distinctions between principal and accessory liability through a contrastive data construction strategy, and fuses case facts with judicial reasoning via label broadcasting and regression alignment. Evaluated on our newly curated IMLJP dataset of intentional injury cases, MMSI significantly outperforms baseline models in role-responsibility disambiguation, achieving higher accuracy while maintaining interpretability and consistency with legal reasoning principles.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Common-Sense Reasoning

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Crime disrupts societal stability, making law essential for balance. In multidefendant cases, assigning responsibility is complex and challenges fairness, requiring precise role differentiation. However, judicial phrasing often obscures the roles of the defendants, hindering effective AI-driven analyses. To address this issue, we incorporate sentencing logic into a pretrained Transformer encoder framework to enhance the intelligent assistance in multidefendant cases while ensuring legal interpretability. Within this framework an oriented masking mechanism clarifies roles and a comparative data construction strategy improves the model's sensitivity to culpability distinctions between principals and accomplices. Predicted guilt labels are further incorporated into a regression model through broadcasting, consolidating crime descriptions and court views. Our proposed masked multistage inference (MMSI) framework, evaluated on the custom IMLJP dataset for intentional injury cases, achieves significant accuracy improvements, outperforming baselines in role-based culpability differentiation. This work offers a robust solution for enhancing intelligent judicial systems, with publicly code available.

Problem

• multidefendant judgment prediction
• role differentiation
• legal interpretability
• culpability distinction
• judicial fairness

Innovation

• Masked Multistage Inference
• Oriented Masking Mechanism
• Role Differentiation
• Explainable AI
• Legal Judgment Prediction