Advancing Legal AI with Graph Neural Networks and Symbolic Logic

This presentation explores advancements in legal AI, emphasizing justice accessibility through interpretable solutions. It introduces Graph Neural Networks and Neural-Symbolic Integration to tackle cost and precision challenges in legal analysis. The research proposes the HRGT model with symbolic constraints, designed to enhance accuracy and scalability. Illustrative examples demonstrate improved multi-hop reasoning and HRGT's efficiency compared to GPT-4, with applications extending to...