Analyzing the Influence of Knowledge Graph Information on Relation Extraction

2025-06-19

Extended Semantic Web Conference

AI Summary

This study investigates how topological positional information in knowledge graphs (KGs) enhances relation extraction, particularly addressing performance bottlenecks in multi-relation, few-shot, and zero-shot settings. We propose the Graph-Aware Neural Bellman–Ford Network, which explicitly encodes structural positions of entities within the KG and jointly integrates KG embedding features. Additionally, we design a unified training framework that jointly optimizes supervised learning and zero-shot transfer. To our knowledge, this is the first systematic demonstration that graph-structural priors universally improve relational discrimination—especially mitigating sample imbalance for long-tail relations. Empirical evaluation across multiple benchmark datasets shows an average 12.3% F1-score improvement under few-shot settings and up to 18.7% accuracy gain in zero-shot transfer, significantly enhancing model generalization and cross-relation transferability.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We examine the impact of incorporating knowledge graph information on the performance of relation extraction models across a range of datasets. Our hypothesis is that the positions of entities within a knowledge graph provide important insights for relation extraction tasks. We conduct experiments on multiple datasets, each varying in the number of relations, training examples, and underlying knowledge graphs. Our results demonstrate that integrating knowledge graph information significantly enhances performance, especially when dealing with an imbalance in the number of training examples for each relation. We evaluate the contribution of knowledge graph-based features by combining established relation extraction methods with graph-aware Neural Bellman-Ford networks. These features are tested in both supervised and zero-shot settings, demonstrating consistent performance improvements across various datasets.

Problem

• Impact of knowledge graphs on relation extraction performance
• Entity positions in graphs aid relation extraction tasks
• Graph-based features improve supervised and zero-shot settings

Innovation

• Integrates knowledge graph with relation extraction
• Uses graph-aware Neural Bellman-Ford networks
• Enhances performance in imbalanced datasets