Perform Like an Engine: A Closed-Loop Neural-Symbolic Learning Framework for Knowledge Graph Inference

TL;DR

This paper introduces EngineKG, a neural-symbolic framework that combines rule learning and KG embeddings to improve knowledge graph inference, achieving better accuracy and interpretability on real-world datasets.

Contribution

The paper presents a novel closed-loop neural-symbolic learning framework that integrates rule pruning and KG embeddings for enhanced knowledge graph inference.

Findings

Outperforms baseline models on link prediction tasks

Improves interpretability of KG inference models

Demonstrates effectiveness across four real-world datasets

Abstract

Knowledge graph (KG) inference aims to address the natural incompleteness of KGs, including rule learning-based and KG embedding (KGE) models. However, the rule learning-based models suffer from low efficiency and generalization while KGE models lack interpretability. To address these challenges, we propose a novel and effective closed-loop neural-symbolic learning framework EngineKG via incorporating our developed KGE and rule learning modules. KGE module exploits symbolic rules and paths to enhance the semantic association between entities and relations for improving KG embeddings and interpretability. A novel rule pruning mechanism is proposed in the rule learning module by leveraging paths as initial candidate rules and employing KG embeddings together with concepts for extracting more high-quality rules. Experimental results on four real-world datasets show that our model outperforms the relevant baselines on link prediction tasks, demonstrating the superiority of our KG inference model in a neural-symbolic learning fashion.