CountTRuCoLa: Rule Learning for Explainable Temporal Knowledge Graph Forecasting

Julia Gastinger, Christian Meilicke, Heiner Stuckenschmidt

Keywords: Temporal Knowledge Graphs, Temporal Graphs, Knowledge Graphs, Temporal Knowledge Graph Forecasting

TL;DR: We introduce a fully explainable method for temporal knowledge graph forecasting based on temporal rules.

Abstract:

We address the task of temporal knowledge graph (TKG) forecasting by introducing a fully interpretable 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 achieves performance that is competitive with state-of-the-art models and consistently outperforms the majority of them, while providing fully interpretable predictions.

Primary Area: learning on graphs and other geometries & topologies

Submission Number: 12131