Link Prediction on N-ary Relational Data Based on Relatedness Evaluation

Guan, Saiping; Jin, Xiaolong; Guo, Jiafeng; Wang, Yuanzhuo none; Cheng, Xueqi

doi:10.1109/tkde.2021.3073483

Cited by 11 publications

(2 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RAM (Liu et al, 2021b) further models the relatedness between different keys and the relatedness between a key and all involved values. NaLP+ (Guan et al, 2021) improves NaLP by considering type information. However, the key-value-based modeling treats all key-value pairs equally and does not distinguish primal triples from qualifiers.…”

Section: Related Workmentioning

confidence: 99%

“…In order to predict links in hyper-relational KGs, pioneering works represent each hyperrelational fact as either an n-tuple in the form of r(e 1 , e 2 , • • • , e n ) (Wen et al, 2016;Zhang et al, 2018;Fatemi et al, 2020;Liu et al, 2020;Abboud et al, 2020) or a set of key-value pairs in the form of {(k i : v i )} m i=1 (Guan et al, 2019(Guan et al, , 2021Liu et al, 2021a). However, these modelings lose key structure information and are incompatible with the RDF-star schema (Arndt et al, 2021) used by modern KGs, where both primal triples and qualifiers constitute the fundamental data structure.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Shrinking Embeddings for Hyper-Relational Knowledge Graphs

Xiong¹,

Nayyeri²,

Pan³

et al. 2023

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

View full text Add to dashboard Cite

Link prediction on knowledge graphs (KGs) has been extensively studied on binary relational KGs, wherein each fact is represented by a triple. A significant amount of important knowledge, however, is represented by hyperrelational facts where each fact is composed of a primal triple and a set of qualifiers comprising a key-value pair that allows for expressing more complicated semantics. Although some recent works have proposed to embed hyper-relational KGs, these methods fail to capture essential inference patterns of hyperrelational facts such as qualifier monotonicity, qualifier implication, and qualifier mutual exclusion, limiting their generalization capability. To unlock this, we present ShrinkE, a geometric hyper-relational KG embedding method aiming to explicitly model these patterns. ShrinkE models the primal triple as a spatial-functional transformation from the head into a relation-specific box. Each qualifier "shrinks" the box to narrow down the possible answer set and, thus, realizes qualifier monotonicity. The spatial relationships between the qualifier boxes allow for modeling core inference patterns of qualifiers such as implication and mutual exclusion. Experimental results demonstrate ShrinkE's superiority on three benchmarks of hyper-relational KGs.

show abstract

Section: Related Workmentioning

confidence: 99%