基于可微知识图谱的多跳知识库问答  

Multi-hop Knowledge Base Question Answering Based on Differentiable Knowledge Graph

作　　者：魏谦强 赵书良[1,2,3] 张思漫 WEI Qianqiang;ZHAO Shuliang;ZHANG Siman(College of Computer and Cyber Security,Hebei Normal University,Shijiazhuang 050024,China;Hebei Provincial Engineering Research Center for Supply China Big Data Analytics&Data Security,Shijiazhuang 050024,China;Hebei Provincial Key Laboratory of Network&Information Security,Shijiazhuang 050024,China;College of Education,Hainan Normal University,Haikou 571158,China)

机构地区：[1]河北师范大学计算机与网络空间安全学院,石家庄050024 [2]供应链大数据分析与数据安全河北省工程研究中心,石家庄050024 [3]河北省网络与信息安全重点实验室,石家庄050024 [4]海南师范大学教育学院,海口571158

出　　处：《计算机科学》2025年第3期295-305,共11页Computer Science

基　　金：国家社会科学基金重大项目(18ZDA200);河北省省级科技计划项目(20370301D,22567606H);河北省引进留学人员项目(C20230339);河北师范大学专项科技基金项目(L2023T03)。

摘　　要：知识库问答是一个具有挑战性的热门研究方向。目前,基于嵌入的方法通过隐式推理得到问题的答案而不能产生完整的推理路径,基于可微知识图谱的模型只需要将问题答案对作为弱监督信号就可以产生可解释的结果。基于可微知识图谱,提出了一个端到端编码器-解码器模型。编码器使用多头注意力机制和LSTM对问题进行细粒度顺序建模,生成能更好地表示问题每一跳语义特征的查询向量;解码器使用前馈神经网络实现问题多跳推理的注意力机制,能更好地表示问题每一跳在整个问题中的权重。所提模型解决了以前粗粒度非顺序建模方法存在的信息丢失问题。在5个数据集MetaQA-1hop,MetaQA-2hop,MetaQA-3hop,WebQSP和CWQ上进行实验,模型分别取得了97.5%,100%,100%,77.8%和51.4%的准确率。消融实验表明,各个模块都对模型整体性能的提高有贡献。Knowledge base question answering(KBQA)is a challenging and popular research direction.Currently,embedding-based methods obtain the answer to a question through implicit reasoning and cannot generate complete reasoning paths.Models based on differentiable knowledge graphs only needs the question-answer pairs as weak supervision signals to generate explainable results.An end-to-end encoder-decoder model based on differentiable knowledge graphs is proposed.The encoder uses multi-head attention mechanism and LSTM to model the fine-grained sequence of questions,generating query vectors that can effectively represent the semantic features of each step of the question.The decoder uses feedforward neural networks to effectively represent the weights of each hop in the entire question.Our model solves the problem of information loss caused by previous coarse-grained and non-sequential modeling methods.The experiments are conducted on five datasets:MetaQA-1hop,MetaQA-2hop,MetaQA-3hop,WebQSP and CWQ,and the model achieves accuracy of 97.5%,100%,100%,77.8% and 51.4%,respectively.The ablation experiment shows that each module contributes to the overall performance improvement of the model.

关 键 词：多跳知识库问答 可微知识图谱 编码器-解码器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]