融合实体特征及多种类注意力机制的领域关系抽取模型  

Domain Relationship Extraction Model Integrating Entity Feature and Multiple Types of Attention Mechanisms

作　　者：王稳 刘远兴 吴湘宁 李文炽 涂雨 张锋 方恒 蔡泽宇 WANG Wen;LIU Yuan-Xing;WU Xiang-Ning;LI Wen-Chi;TU Yu;ZHANG Feng;FANG Heng;CAI Ze-Yu(School of Computer Science,China University of Geosciences,Wuhan 430078,China)

机构地区：[1]中国地质大学(武汉)计算机学院,武汉430078

出　　处：《计算机系统应用》2024年第4期202-208,共7页Computer Systems & Applications

基　　金：国家自然科学基金(U21A2013);智能地学信息处理湖北省重点实验室开放基金(KLIGIP-2018B14)。

摘　　要：基于远程监督的关系抽取方法可以明显地减少人工标注数据集的成本,已经被广泛应用于领域知识图谱的构建任务中.然而,现有的远程监督关系抽取方法领域针对性不强,同时也忽略了对领域实体特征信息的利用.为了解决上述问题,提出了一种融合实体特征和多种类注意力机制的关系抽取模型PCNN-EFMA.模型采用远程监督和多实例技术,不再受限于人工标注.同时,为了减少远程监督中噪声的影响,模型使用了句子注意力和包间注意力这两类注意力,并在词嵌入层和句子注意力中融合实体特征信息,增强了模型的特征选择能力.实验表明,该模型在领域数据集上的PR曲线更好,并在P@N上的平均准确率优于PCNN-ATT模型.The relationship extraction method based on remote supervision can cut the cost of labor-based annotated datasets and has been widely used in the construction of the domain knowledge graph.However,the existing remote supervised relationship extraction methods are not domain-specific and also neglect the utilization of domain entity feature information.To solve the above problems,this study proposes a relationship extraction model PCNN-EFMA that integrates entity features and multiple types of attention mechanisms.The model adopts remote supervision and multiinstance technology,no longer limited by labor-based annotation.At the same time,to reduce the impact of noise in remote supervision,the model uses two types of attention:sentence attention and inter-packet attention.In addition,it integrates entity feature information in the word embedding layer and sentence attention,enhancing the model’s feature selection ability.Experiments show that the PR curve of this model is better on the domain dataset,and its average accuracy on P@N is better than that of the PCNN-ATT model.

关 键 词：关系抽取 知识图谱 注意力机制 实体特征 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]