知识融合中的实体对齐方法简介
首发时间:2020-04-07
摘要:随着近年来知识图谱技术的发展,以图的形式来表示知识已经成为一种大热趋势。但对于领域知识图谱来说,存在信息覆盖率低,知识描述和知识质量低,知识异质性强等问题。因此知识图谱中的知识融合技术引发了越来越多学者的关注,而实体对齐技术便首当其冲。实体对齐是指将存在于不同的知识图谱中,但指向现实世界中相同客观对象的实体链接起来。本文首先介绍了实体对齐任务的基本概念,接着介绍了主流的实体对齐方法,主要包括有监督方法和无监督方法,然后说明了实体对齐实验中主要的数据集和工具、主流的评测方法和相关方法的实验结果等,最后总结了全文,并对未来实体对齐技术的发展提出了展望。
关键词: 人工智能 实体对齐 知识图谱 知识表示 图卷积网络。
For information in English, please click here
A Survey of Entity Alignment Methods in Knowledge Fusion
Abstract:With the development of knowledge graph technology in recent years, it has become a hot trend to represent knowledge in the form of graphs. But for the domain knowledge map, there are problems such as low information coverage, low knowledge description and quality, and strong knowledge heterogeneity. Therefore, the knowledge fusion technology in the knowledge map has attracted more and more scholars\' attention, and the entity alignment technology is the first to bear the brunt. Entity alignment refers to linking entities that exist in different knowledge maps but point to the same objective object in the real world. This article first introduces the basic concepts of the entity alignment task, and then introduces the mainstream entity alignment methods, mainly including supervised and unsupervised methods, and then explains the main data sets and tools, mainstream evaluation methods and related methods in entity alignment experiments. The experimental results of the method, etc., finally summarized the full text, and proposed the development of future entity alignment technology.
Keywords: artificial intelligence entity alignment knowledge graph knowledge representation graph convolutional network.
引用
No.****
动态公开评议
共计0人参与
勘误表
知识融合中的实体对齐方法简介
评论
全部评论0/1000