CountTRuCoLa: Rule Confidence Learning for Temporal Knowledge Graph Forecasting

2025-09-11

AI Summary

To address the lack of interpretability in temporal knowledge graph (TKG) forecasting, this paper proposes the first fully interpretable temporal rule learning framework. Methodologically, it introduces and mines four categories of atomic temporal rules—e.g., “if relation (r_1) holds at time (t), then (r_2) is highly likely to hold at (t+1)”—and innovatively designs a dual-factor confidence scoring mechanism that jointly models temporal novelty (recency) and statistical robustness (support frequency). Rule discovery and weighted evaluation are achieved efficiently via recursive fact analysis. Extensive experiments across nine standard TKG benchmarks demonstrate that our approach consistently outperforms eight state-of-the-art models and two baselines in predictive accuracy, while simultaneously delivering human-understandable, logically traceable prediction justifications—achieving both high performance and strong interpretability.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Planning, Routing, and Scheduling: Temporal Planning

Application Category

• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We address the task of temporal knowledge graph (TKG) forecasting by introducing a fully explainable method based on temporal rules. Motivated by recent work proposing a strong baseline using recurrent facts, our approach learns four simple types of rules with a confidence function that considers both recency and frequency. Evaluated on nine datasets, our method matches or surpasses the performance of eight state-of-the-art models and two baselines, while providing fully interpretable predictions.

Problem

• Develops explainable temporal knowledge graph forecasting method
• Learns rule confidence using recency and frequency factors
• Evaluates against state-of-the-art models on nine datasets

Innovation

• Temporal rules for explainable forecasting
• Confidence function using recency and frequency
• Four simple rule types for interpretable predictions