arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2009.08694cs.CLcs.AI
收藏

RECON:在图神经网络中使用知识图谱上下文的关系抽取

RECON: Relation Extraction using Knowledge Graph Context in a Graph Neural Network

IIT Hyderabad(印度理工学院海得拉巴分校);Zerotha Research(Zerotha研究公司);RWTH Aachen(亚琛工业大学);Cerence GmbH(赛轮思有限公司);Fraunhofer IAIS(弗劳恩霍夫智能分析与信息系统研究所);University of Dayton(戴顿大学);Goldman Sachs(高盛集团)

Anson Bastos, Abhishek Nadgeri, Kuldeep Singh, Isaiah Onando Mulang', Saeedeh Shekarpour, Johannes Hoffart, Manohar Kaul

摘要:

在本文中,我们提出了一种新的方法,名为RECON,自动识别一个句子中的关系(间接关系抽取),并对齐到一个知识图谱(KG)。RECON使用图神经网络来学习句子以及KG中存储的事实的表示,从而提高整体提取质量。这些事实,包括实体属性(标签、别名、描述、实例)和事实三元组,尚未在现有技术的方法中共同使用。我们评估了各种形式的代表KG上下文对RECON性能的影响。两个标准的关系抽取数据集的实证评估表明,RECON显着优于所有国家的最先进的方法在纽约时报Freebase和维基数据集。RECON在Wikidata数据集上报告了87.23 F1评分(与基线82.29相比),而在NYT Freebase上,报告的值为87.5(P@10)和74.1(P@30),而之前的基线评分为81.3(P@10)和63.1(P@30)。

英文摘要:

In this paper, we present a novel method named RECON, that automatically identifies relations in a sentence (sentential relation extraction) and aligns to a knowledge graph (KG). RECON uses a graph neural network to learn representations of both the sentence as well as facts stored in a KG, improving the overall extraction quality. These facts, including entity attributes (label, alias, description, instance-of) and factual triples, have not been collectively used in the state of the art methods. We evaluate the effect of various forms of representing the KG context on the performance of RECON. The empirical evaluation on two standard relation extraction datasets shows that RECON significantly outperforms all state of the art methods on NYT Freebase and Wikidata datasets. RECON reports 87.23 F1 score (Vs 82.29 baseline) on Wikidata dataset whereas on NYT Freebase, reported values are 87.5(P@10) and 74.1(P@30) compared to the previous baseline scores of 81.3(P@10) and 63.1(P@30).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑