基于深度学习的领域本体构建研究——以汽车领域为例  

Research on Domain Ontology Hierarchical Construction Method Based on Deep Learning ——Taking the Automobile Field as an Example

作　　者：陈晓燕[1] 贾珊 何有世[1] Chen Xiaoyan;Jia Shan;He Youshi(School of Management,Jiangsu University,Zhenjiang 212013)

机构地区：[1]江苏大学管理学院

出　　处：《情报杂志》2020年第1期174-178,共5页Journal of Intelligence

基　　金：国家社会科学基金项目“基于领域本体的企业产品网络口碑危机预警研究”(编号:16BGL088);江苏省六大人才高峰项目“城市交通系统节能减排机制和优化设计方法研究”(编号:JXQC-037)研究成果之一

摘　　要：[目的/意义]结合深度学习技术和领域本体研究,提出了基于深度学习的领域本体构建方法,从概念抽取、层级关系构建、属性抽取三个模块,结合人工干预搭建起本体架构。[方法/过程]利用深度神经网络实现领域本体概念识别,以BIRCH聚类算法划分本体层级关系,采用卷积神经网络结合远程监督抽取实体关系模型,来抽取领域本体属性。[结果/结论]以汽车领域评论语料为数据源,搭建起汽车领域本体架构,为领域本体构建提供了一种新思路。[Purpose/Significance]Combining deep learning technology and domain ontology research,this paper proposes a domain ontology construction method based on deep learning.The ontology architecture is built from three modules:concept extraction,hierarchical relationship construction and attribute extraction.[Method/Process]The deep domain neural network is used to realize the domain ontology concept recognition.The BIRCH clustering algorithm is used to construct the ontology hierarchical relationship.The convolutional neural network is combined with the remote supervision to extract the entity relationship model and to extract the domain ontology attributes.[Result/Conclusion]Taking the automotive domain commentary corpus as the data source,the auto domain ontology architecture is built,which provides a new idea for the domain ontology automation construction.

关 键 词：领域本体 深度学习 深度神经网络 BIRCH算法 

分 类 号：C931[经济管理—管理学]