Connecting Embeddings for Knowledge Graph Entity Typing

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6419–6428 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6419 Connecting Embeddings for Knowledge Graph Entity Typing Yu Zhao1,∗, Anxiang Zhang2,*, Ruobing Xie3, Kang Liu4,5, Xiaojie Wang6 1Fintech Innovation Center, School of Economic Information Engineering, Southwestern University of Finance and Economics, Chengdu, China 2School of Computer Science, Carnegie Mellon University, Pittsburgh, USA 3WeChat Search Application Department, Tencent, Beijing, China 4National Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China 5University of Chinese Academy of Sciences, Beijing, 100049, China 6 School of Computer Science, Beijing University of Posts and Telecommunications, Beijing, China Abstract Knowledge graph (KG) entity typing aims at inferring possible missing entity type in- stances in KG, which is a very significant but still under-explored subtask of knowledge graph completion. In this paper, we pro- pose a novel approach for KG entity typ- ing which is trained by jointly utilizing lo- cal typing knowledge from existing entity type assertions and global triple knowledge from KGs. Specifically, we present two distinct knowledge-driven effective mechanisms of en- tity type inference. Accordingly, we build two novel embedding models to realize the mech- anisms. Afterward, a joint model with them is used to infer missing entity type instances, which favors inferences that agree with both entity type instances and triple knowledge in KGs. Experimental results on two real-world datasets (Freebase and YAGO) demonstrate the effectiveness of our proposed mechanisms and models for improving KG entity typing. The source code and data of this paper can be obtained from: https://github.com/ Adam1679/ConnectE 1 Introduction The past decade has witnessed great thrive in build- ing web-scale knowledge graphs (KGs), such as Freebase (Bollacker et al., 2008), YAGO (Suchanek et al., 2007), Google Knowledge Graph (Dong et al., 2014), which usually consists of a huge amount of triples in the form of (head entity, rela- tion, tail entity) (denoted (e, r, ˜e)). KGs usually suf- fer from incompleteness and miss important facts, jeopardizing their usefulness in downstream tasks such as question answering (Elsahar et al., 2018), semantic parsing (Berant et al., 2013), relation clas- sification (Zeng et al., 2014). Hence, the task of ∗ Equal Contribution. Corresponding author: Y. Zhao (zhaoyu@swufe.edu.cn). Figure 1: Effective mechanisms of entity type inference with local typing knowledge and global triple knowl- edge. knowledge graph completion (KGC, i.e. complet- ing knowledge graph entries) is extremely signifi- cant and attracts wide attention. This paper concentrates on KG entity typing, i.e. inferring missing entity type instances in KGs, which is an important sub-problem of KGC. En- tity type instances, each of which is in the formed of (entity, entity type) (denoted (e, t)), are essen- tial entries of KGs and widely used in many NLP tasks such as relation extraction (Zhang et al., 2018; Jain et al., 2018), coreference resolution (Hajishirzi et al., 2013), entity linking (Gupta et al., 2017). Most previous works of KGC focus on inferring missing entities and relationships (Bordes …

Content truncated to 500 words