Message Passing for Hyper-Relational Knowledge Graphs

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 7346–7359, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 7346 Message Passing for Hyper-Relational Knowledge Graphs Mikhail Galkin1,2, Priyansh Trivedi2, Gaurav Maheshwari2, Ricardo Usbeck2, Jens Lehmann2,3 1TU Dresden, 2Fraunhofer IAIS, 3University of Bonn {mikhail.galkin, priyansh.trivedi, gaurav.maheshwari, ricardo.usbeck, jens.lehmann}@iais.fraunhofer.de Abstract Hyper-relational knowledge graphs (KGs) (e.g., Wikidata) enable associating additional key-value pairs along with the main triple to disambiguate, or restrict the validity of a fact. In this work, we propose a message passing based graph encoder - STARE capable of modeling such hyper-relational KGs. Un- like existing approaches, STARE can encode an arbitrary number of additional information (qualifiers) along with the main triple while keeping the semantic roles of qualifiers and triples intact. We also demonstrate that exist- ing benchmarks for evaluating link prediction (LP) performance on hyper-relational KGs suf- fer from fundamental flaws and thus develop a new Wikidata-based dataset - WD50K. Our ex- periments demonstrate that STARE based LP model outperforms existing approaches across multiple benchmarks. We also confirm that leveraging qualifiers is vital for link prediction with gains up to 25 MRR points compared to triple-based representations. 1 Introduction The task of link prediction over knowledge graphs (KGs) has seen a wide variety of advances over the years (Ji et al., 2020). The objective of this task is to predict new links between entities in the graph based on the existing ones. A majority of these approaches are designed to work over triple- based KGs, where facts are represented as binary relations between entities. This data model, how- ever, doesn’t allow for an intuitive representation of facts with additional information. For instance, in Fig. 1.A, it is non-trivial to add information which can help disambiguate whether the two universities attended by Albert Einstein awarded him with the same degree. This additional information can be provided in the form of key-value restrictions over instances of educated at educated at Albert Einstein Q937 ETH Zurich University of Zurich Academic degree (P512): Bachelor (Q787674) Academic major (P812): Mathematics (Q853077) Academic degree (P512): Doctorate (Q849697) Academic major (P812): Physics (Q413) A. Triple-based facts B. Hyper-relational facts P69 P69 Q206702 Q11942 educated at educated at Albert Einstein Q937 ETH Zurich University of Zurich P69 P69 Q206702 Q11942 Figure 1: A comparison of triple-based and hyper- relational facts. binary relations between entities in recent knowl- edge graph models (Vrandecic and Kr¨otzsch, 2014; Pellissier-Tanon et al., 2020; Ismayilov et al., 2018). Such restrictions are known as qualifiers in the Wikidata statement model (Vrandecic and Kr¨otzsch, 2014) or triple metadata in RDF* (Har- tig, 2017) and RDF reification approaches (Frey et al., 2019). These complex facts with qualifiers can be represented as hyper-relational facts (See Sec. 3). In our example (Fig. 1.B), hyper-relational facts allow to observe that Albert Einstein obtained different degrees at those universities. Existing representation learning approaches for such graphs largely treat a hyper-relational fact as an n-ary (n>2) composed relation (e.g., educatedAt academicDegree) (Zhang et al., 2018; Liu et …

Content truncated to 500 words