DHGE: Dual-View Hyper-Relational Knowledge Graph Embedding for Link Prediction and Entity Typing

Authors

  • Haoran Luo School of Computer Science, Beijing University of Posts and Telecommunications
  • Haihong E School of Computer Science, Beijing University of Posts and Telecommunications
  • Ling Tan School of Computer Science, Beijing University of Posts and Telecommunications
  • Gengxian Zhou School of Computer Science, Beijing University of Posts and Telecommunications
  • Tianyu Yao School of Computer Science, Beijing University of Posts and Telecommunications
  • Kaiyang Wan School of Computer Science, Beijing University of Posts and Telecommunications

DOI:

https://doi.org/10.1609/aaai.v37i5.25795

Keywords:

KRR: Ontologies and Semantic Web, DMKM: Other Foundations of Data Mining & Knowledge Management, DMKM: Semantic Web, APP: Healthcare, Medicine & Wellness, KRR: Knowledge Acquisition, KRR: Knowledge Engineering, KRR: Knowledge Representation Languages, SNLP: Information Extraction, SNLP: Interpretability & Analysis of NLP Models, SNLP: Ontology Induction From Text, SNLP: Sentence-Level Semantics and Textual Inference

Abstract

In the field of representation learning on knowledge graphs (KGs), a hyper-relational fact consists of a main triple and several auxiliary attribute-value descriptions, which is considered more comprehensive and specific than a triple-based fact. However, currently available hyper-relational KG embedding methods in a single view are limited in application because they weaken the hierarchical structure that represents the affiliation between entities. To overcome this limitation, we propose a dual-view hyper-relational KG structure (DH-KG) that contains a hyper-relational instance view for entities and a hyper-relational ontology view for concepts that are abstracted hierarchically from the entities. This paper defines link prediction and entity typing tasks on DH-KG for the first time and constructs two DH-KG datasets, JW44K-6K, extracted from Wikidata, and HTDM based on medical data. Furthermore, we propose DHGE, a DH-KG embedding model based on GRAN encoders, HGNNs, and joint learning. DHGE outperforms baseline models on DH-KG, according to experimental results. Finally, we provide an example of how this technology can be used to treat hypertension. Our model and new datasets are publicly available.

Downloads

Published

2023-06-26

How to Cite

Luo, H., E, H., Tan, L., Zhou, G., Yao, T., & Wan, K. (2023). DHGE: Dual-View Hyper-Relational Knowledge Graph Embedding for Link Prediction and Entity Typing. Proceedings of the AAAI Conference on Artificial Intelligence, 37(5), 6467-6474. https://doi.org/10.1609/aaai.v37i5.25795

Issue

Section

AAAI Technical Track on Knowledge Representation and Reasoning