CHAINSFORMER: Numerical Reasoning on Knowledge Graphs from a Chain Perspective

2025-04-19

AI Summary

Existing knowledge graph (KG) numerical reasoning methods—such as graph neural networks (GNNs) and knowledge graph embedding (KGE) approaches—rely on local neighborhood aggregation and implicit vector representations, limiting their ability to model long-range logical paths, resulting in insufficient reasoning depth and poor interpretability. To address this, we propose RA-Chains, the first framework to explicitly model relation-attribute logical chains (RA-Chains) for multi-hop numerical reasoning. We design a hyperspherical affinity scoring mechanism operating in hyperbolic space to robustly filter chains under noise, and introduce an attention-driven numerical reasoner that jointly optimizes accuracy and transparency. Extensive experiments on multiple KG numerical reasoning benchmarks demonstrate significant improvements over state-of-the-art methods, with up to a 20.0% absolute gain in accuracy. The source code is publicly available.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs 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 query analysis, representation and understanding

Abstract

Reasoning over Knowledge Graphs (KGs) plays a pivotal role in knowledge graph completion or question answering systems, providing richer and more accurate triples and attributes. As numerical attributes become increasingly essential in characterizing entities and relations in KGs, the ability to reason over these attributes has gained significant importance. Existing graph-based methods such as Graph Neural Networks (GNNs) and Knowledge Graph Embeddings (KGEs), primarily focus on aggregating homogeneous local neighbors and implicitly embedding diverse triples. However, these approaches often fail to fully leverage the potential of logical paths within the graph, limiting their effectiveness in exploiting the reasoning process. To address these limitations, we propose ChainsFormer, a novel chain-based framework designed to support numerical reasoning. Chainsformer not only explicitly constructs logical chains but also expands the reasoning depth to multiple hops. Specially, we introduces Relation-Attribute Chains (RA-Chains), a specialized logic chain, to model sequential reasoning patterns. ChainsFormer captures the step-by-step nature of multi-hop reasoning along RA-Chains by employing sequential in-context learning. To mitigate the impact of noisy chains, we propose a hyperbolic affinity scoring mechanism that selects relevant logic chains in a variable-resolution space. Furthermore, ChainsFormer incorporates an attention-based numerical reasoner to identify critical reasoning paths, enhancing both reasoning accuracy and transparency. Experimental results demonstrate that ChainsFormer significantly outperforms state-of-the-art methods, achieving up to a 20.0% improvement in performance. The implementations are available at https://github.com/zhaodazhuang2333/ChainsFormer.

Problem

• Enables numerical reasoning on knowledge graphs via logical chains
• Addresses limitations of GNNs/KGEs in exploiting multi-hop reasoning paths
• Improves accuracy by filtering noisy chains with hyperbolic scoring

Innovation

• Explicitly constructs logical chains for reasoning
• Uses hyperbolic affinity scoring for noise reduction
• Employs attention-based numerical reasoner for accuracy