Abstract:Knowledge graph reasoning (KGR) aims to infer new knowledge or detect noises, which is essential for improving the quality of knowledge graphs. Recently, various KGR techniques, such as symbolic- and embedding-based methods, have been proposed and shown strong reasoning ability. Symbolic-based reasoning methods infer missing triples according to predefined rules or ontologies. Although rules and axioms have proven effective, it is difficult to obtain them. Embedding-based reasoning methods represent entities and relations as vectors, and complete KGs via vector computation. However, they mainly rely on structural information and ignore implicit axiom information not predefined in KGs but can be reflected in data. That is, each correct triple is also a logically consistent triple and satisfies all axioms. In this paper, we propose a novel NeuRal Axiom Network (NeuRAN) framework that combines explicit structural and implicit axiom information without introducing additional ontologies. Specifically, the framework consists of a KG embedding module that preserves the semantics of triples and five axiom modules that encode five kinds of implicit axioms. These axioms correspond to five typical object property expression axioms defined in OWL2, including ObjectPropertyDomain, ObjectPropertyRange, DisjointObjectProperties, IrreflexiveObjectProperty and AsymmetricObjectProperty. The KG embedding module and axiom modules compute the scores that the triple conforms to the semantics and the corresponding axioms, respectively. Compared with KG embedding models and CKRL, our method achieves comparable performance on noise detection and triple classification and achieves significant performance on link prediction. Compared with TransE and TransH, our method improves the link prediction performance on the Hits@1 metric by 22.0% and 20.8% on WN18RR-10% dataset, respectively.

Infer Latent Privacy for Attribute Network in Knowledge Graph

Devil in Disguise: Breaching Graph Neural Networks Privacy through Infiltration

De-anonymizing Social Networks and Inferring Private Attributes Using Knowledge Graphs

Social Network De-Anonymization and Privacy Inference with Knowledge Graph Model

Inferring Private Attributes Based on Graph Convolutional Neural Network in Social Networks

Multi-task Neural Network for Non-discrete Attribute Prediction in Knowledge Graphs

A Triple-Branch Neural Network for Knowledge Graph Embedding.

Privacy Inference on Knowledge Graphs: Hardness and Approximation

Neural Axiom Network for Knowledge Graph Reasoning

A Privacy Inference Model Based on Attribute Graph.

A Privacy Preserving Graph Neural Networks Framework by Protecting User’s Attributes

Privacy-Preserving Representation Learning on Graphs

Quantifying Privacy Leakage in Graph Embedding

Quantifying and Defending against Privacy Threats on Federated Knowledge Graph Embedding

Differentially Private Federated Knowledge Graphs Embedding

KTAT: A Complex Embedding Model of Knowledge Graph Integrating Type Information and Attention Mechanism

Application of Graph Neural Network and Feature Information Enhancement in Relation Inference of Sparse Knowledge Graph

Graph-based neural networks for explainable image privacy inference

Triple Trustworthiness Measurement for Knowledge Graph

Privacy-Preserving Representation Learning on Graphs: A Mutual Information Perspective

Adaptive Attention-based Knowledge Graph Completion