Rule-based data augmentation for knowledge graph embedding

Rule-based data augmentation for knowledge graph embedding
复制标题

用于知识图嵌入的基于规则的数据增强

DOI:
10.1016/j.aiopen.2021.09.003
复制
发表时间:
2021
期刊:
AI Open
影响因子:
--
通讯作者:
Wei Hu
Wei Hu
中科院分区:
其他
文献类型:
--
作者:
Guangyao Li;Zequn Sun;Lei Qian;Qiang Guo;Wei Hu

文献摘要

参考文献

相似文献

知识图(KG)嵌入模型存在观测事实的不完全性问题。不同于现有的解决方案,包括额外的信息或采用表达和复杂的嵌入技术,我们建议增加KG迭代挖掘逻辑规则从观察到的事实,然后使用规则来生成新的关系三元组。随着新的增强三元组的到来,我们逐渐训练KG嵌入,并利用嵌入来验证这些新的三元组。为了保证增强数据的质量,我们在验证过程中过滤掉基于传播机制的噪声三元组。挖掘出的规则和规则基础是人类可以理解的,并且可以使增强过程可靠。我们的KG增强框架适用于任何KG嵌入模型,无需修改其嵌入技术。我们在两个流行的基于嵌入的任务上的实验(即,实体对齐和链接预测)表明,所提出的框架可以带来显着的改进,现有的KG嵌入模型在大多数基准数据集。
Knowledge graph (KG) embedding models suffer from the incompleteness issue of observed facts. Different from existing solutions that incorporate additional information or employ expressive and complex embedding techniques, we propose to augment KGs by iteratively mining logical rules from the observed facts and then using the rules to generate new relational triples. We incrementally train KG embeddings with the coming of new augmented triples, and leverage the embeddings to validate these new triples. To guarantee the quality of the augmented data, we filter out the noisy triples based on a propagation mechanism during the validation. The mined rules and rule groundings are human-understandable, and can make the augmentation procedure reliable. Our KG augmentation framework is applicable to any KG embedding models with no need to modify their embedding techniques. Our experiments on two popular embedding-based tasks (i.e., entity alignment and link prediction) show that the proposed framework can bring significant improvement to existing KG embedding models on most benchmark datasets.
DOI: 10.14778/3407790.3407828
发表时间: 2020-03
影响因子: 2.5
作者:
Zequn Sun;Qingheng Zhang;Wei Hu;Chengming Wang;Muhao Chen;F. Akrami;Chengkai Li
通讯作者: Zequn Sun;Qingheng Zhang;Wei Hu;Chengming Wang;Muhao Chen;F. Akrami;Chengkai Li
DOI: 10.1145/3424672
发表时间: 2021-04-01
影响因子: 3.6
作者:
Rossi, Andrea;Barbosa, Denilson;Merialdo, Paolo
通讯作者: Merialdo, Paolo
DOI: 10.1109/tnnls.2021.3070843
发表时间: 2021-04-24
影响因子: 10.4
作者:
Ji, Shaoxiong;Pan, Shirui;Yu, Philip S.
通讯作者: Yu, Philip S.
DOI: --
发表时间: 2017-12
期刊: ArXiv
影响因子: --
作者:
Luis Perez;Jason Wang
通讯作者: Luis Perez;Jason Wang