基于语义增强的多特征融合小样本关系抽取  

Multi-feature fusion and few-shot relation extractionbased on semantic enhancement

作　　者：潘理虎[1] 刘云 谢斌红[1] 张英俊[1] Pan Lihu;Liu Yun;Xie Binhong;Zhang Yingjun(College of Computer Science&Technology,Taiyuan University of Science&Technology,Taiyuan 030024,China)

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《计算机应用研究》2022年第6期1663-1667,共5页Application Research of Computers

基　　金：山西省自然科学基金资助项目(201901D111258);山西省中科院科技合作项目(20141101001);山西省重点研究计划资助项目(201603D121031);山西省应用基础研究项目(201801D221179)。

摘　　要：关系抽取是自然语言处理和知识图谱构建的一项关键任务,现有小样本关系抽取方法无法有效获取和充分利用更多的文本语义信息,为此给出一个基于语义增强的多特征融合关系抽取方法(SMPC)并应用于小样本任务中。该方法构建了一个融合位置、词性和句法依存等信息的分段卷积神经网络最大化表示语义特征,并从维基百科中抽取细粒度语义信息,将其融入词嵌入,提高模型上下文共性学习。在两种场景下对不同基线方法进行实验,分别取得最高4%和10%的准确率提升,证明了该方法的有效性。Entity relationship extraction is a key task of natural language processing and knowledge graph construction.The existing few-shot relation extraction methods can’t effectively obtain and make full use of more text semantic information.For this reason,this paper proposed a semantic enhancement-based multi-feature fusion relationship extraction method(SMPC),and applied it to few-shot tasks.The method constructed a piecewise convolutional neural network that integrated information such as position,part of speech,and syntactic dependence to maximize semantic features.The model extracted fine-grained semantic information from Wikipedia and integrated it into word embedding to improve the common learning of model context.This paper made experiments on different baseline methods in two scenarios.It achieves a maximum accuracy improvement of 4%and 10%,and proves the effectiveness of the method.

关 键 词：关系抽取 小样本学习 细粒度语义 特征融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]