TA-KAND: Two-stage Attention Triple Enhancement and U-KAN based Diffusion For Few-shot Knowledge Graph Completion

2025-12-13

AI Summary

To address the few-shot knowledge graph (KG) completion challenge posed by long-tailed relations, this paper proposes a generative modeling framework. First, it introduces a two-stage attention-based triplet augmenter that explicitly captures local neighborhood structure and semantic contrast signals. Second, it pioneers the integration of a U-shaped Kolmogorov–Arnold Network (U-KAN) into a denoising diffusion probabilistic model (DDPM), enabling joint modeling of topological structure and contrastive distribution. The method unifies graph neural networks with contrastive learning, eliminating alignment bias and distribution mismatch inherent in prior approaches. Evaluated on FB15k-237 and NELL-995 under the 5-shot setting, the method achieves a new state-of-the-art mean reciprocal rank (MRR), improving by 4.2% over the previous best, and significantly outperforms existing few-shot KG completion methods.

Technology Category

• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

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

Abstract

Knowledge Graphs (KGs), thanks to their concise and efficient triple-based structure, have been widely applied in intelligent question answering, recommender systems and other domains. However, the heterogeneous and multifaceted nature of real-world data inevitably renders the distribution of relations long-tailed, making it crucial to complete missing facts with limited samples. Previous studies mainly based on metric matching or meta learning, yet they either fail to fully exploit neighborhood information in graph or overlook the distributional characteristics of contrastive signals. In this paper, we re-examine the problem from a perspective of generative representation and propose a few-shot knowledge graph completion framework that integrates two-stage attention triple enhancer with U-KAN based diffusion model. Extensive experiments on two public datasets show that our method achieve new state-of-the-art results.

Problem

• Completes missing facts in knowledge graphs with limited samples
• Addresses long-tailed relation distribution in heterogeneous real-world data
• Enhances few-shot learning using generative representation and diffusion models

Innovation

• Two-stage attention triple enhancer for neighborhood exploitation
• U-KAN based diffusion model for generative representation
• Integration of enhancer and diffusion for few-shot completion