OAGknow: Self-Supervised Learning for Linking Knowledge Graphs

doi:10.1109/TKDE.2021.3090830

OAGknow: Self-Supervised Learning for Linking Knowledge Graphs

2021-01-01

4

PRE

AI

DOI：10.1109/TKDE.2021.3090830

Original

Original request for help

Save

Abstract

En 中文

propose a self-supervised embedding learning framework-SelfLinKG-to link concepts in heterogeneous knowledge graphs. Without any labeled data, SelfLinKG can achieve competitive performance against its supervised counterpart, and significantly outperforms state-of-the-art unsupervised methods by 26%-50% under linear classification protocol. The essential components of SelfLinKG are local attention-based encoding and momentum contrastive learning. The former aims to learn the graph representation using an attention network, while the latter is to learn a self-supervised model across knowledge graphs using contrastive learning. SelfLinKG has been deployed to build the the new version, called OAG(know) of Open Academic Graph (OAG). All data and codes are publicly available.

Keywords:

Encyclopedias

Internet

Electronic publishing

Knowledge based systems

Taxonomy

Training

Encoding

Concept linking

self-supervised learning

contrastive learning

knowledge base

OAGknow: Self-Supervised Learning for Linking Knowledge Graphs

Abstract

AI Summary

Journal

Organization