陕西师范大学学报（自然科学版）2026，Vol.54Issue(5)：1-10,10.DOI:10.15983/j.cnki.jsnu.2026502
融合三段论推理和知识增强的司法问答方法
Method for judicial question-answering by combining syllogistic reasoning and knowledge augmentation
摘要
Abstract
Due to the limitations of large language models in deeply understanding legal knowledge and grasping the logical connections between legal provisions and judicial issues,their performance in the field of judicial question-answering is constrained.To address these challenges,a judicial question-answering method that integrates syllogistic reasoning and knowledge enhancement is proposed.The method first incorporates legal provisions as an external knowledge base to compensate for the model's deficiencies in judicial domain knowledge.Then,it constructs a chain of thought based on judicial syllogism to strengthen the model's ability to discern the inherent logical relationships between judicial issues and legal provisions.Finally,the large language model combines legal provisions and judicial syllogism for reasoning,yielding more accurate and legally consistent answers.Experimental results demonstrate that the proposed method improves performance by 10.1%and 12.4%on judicial examination multiple-choice questions and judicial consultation question-answering datasets,respectively,compared to baseline models,validating its effectiveness.关键词
大语言模型/知识增强/司法问答/法律推理/司法三段论Key words
large language models/knowledge enhancement/judicial question answering/legal reasoning/judicial syllogism分类
信息技术与安全科学引用本文复制引用
肖立东,林川,黄瑞章,秦永彬,陈艳平,周裕林..融合三段论推理和知识增强的司法问答方法[J].陕西师范大学学报（自然科学版）,2026,54(5):1-10,10.基金项目
国家重点研发计划(2023YFC3304500) （2023YFC3304500）
贵州省科学技术基金重点资助项目(黔科合重大专项字[2024]003) （黔科合重大专项字[2024]003）
贵州省科技支撑计划项目(黔科合支撑[2023]一般448) （黔科合支撑[2023]一般448）
