Abstract:The rapid development of natural language processing (NLP) technologies has enabled the emergence of legal intelligence assistance systems, with legal charge prediction (LCP) being a critical technology. The automatic LCP aims to determine the final charges based on fact descriptions of criminal cases. LCP assists human judges in managing workloads and improving efficiency, provides accessible legal guidance for individuals, and supports enterprises in litigation financing and compliance monitoring. However, distinguishing between confusing charges in real-world judicial practice remains a significant challenge. Most exist works cannot effectively capture complex relationships and discern subtle differences in fact descriptions while ignoring the legal schematic knowledge. In order to improve confusing LCP performance, we propose a novel knowledge-aware model for legal charge prediction that leverages Graph Neural Networks (GNNs) to capture complex relationships within criminal case descriptions. Specifically, the model constructs structural and semantic graphs from fact descriptions and integrates information from both through a dual-graph interaction process. A legal knowledge transformer generates key knowledge representations at schema and charge levels, while a knowledge matching network incorporates hierarchical charge knowledge into facts. Besides, we also propose two real-world datasets namely Criminal-All and Criminal-Confusing, containing 203 different charges and 86 confusing charges, respectively. To the best of our knowledge, these datasets are the first well-organized datasets for confusing LCP task. Experimental results demonstrate that the proposed model outperforms baselines and significantly improves the distinction of confusing charges, providing valuable support for intelligent legal judgment systems.

BLHNN: A Novel Charge Prediction Model Based on Bi-Attention LSTM-CNN Hybrid Neural Network

Charge Prediction with Legal Attention

Learning to Predict Charges for Criminal Cases with Legal Basis

A hybrid neural network model for predicting charges by incorporating legal knowledge

SIM-GCN: similarity graph convolutional networks for charges prediction

SLBCNN: A Improved Deep Learning Model for Few-Shot Charge Prediction

Knowledge-aware Method for Confusing Charge Prediction

ADAN: an Intelligent Approach Based on Attentive Neural Network and Relevant Law Articles for Charge Prediction.

Learning To Predict Charges For Judgment With Legal Graph

Knowledge-enhanced model with dual-graph interaction for confusing legal charge prediction

Mulan: A Multiple Residual Article-Wise Attention Network for Legal Judgment Prediction

Knowledge-Enriched Multi-Cross Attention Network for Legal Judgment Prediction

Interpretable Charge Prediction for Criminal Cases with Dynamic Rationale Attention

Legal Charge Prediction Via Bilinear Attention Network

Few-Shot Charge Prediction with Discriminative Legal Attributes.

MANN: A Multichannel Attentive Neural Network for Legal Judgment Prediction

Learning to Predict Charges for Legal Judgment Via Self-Attentive Capsule Network.

Element-Aware Legal Judgment Prediction for Criminal Cases with Confusing Charges

Interpretable Charge Prediction for Legal Cases Based on Interdependent Legal Information

Charge Prediction for Multi-defendant Cases with Multi-scale Attention.