TeRoR: Decoupled Temporal Rotation with Relational Circular Region for Temporal Knowledge Graph Embedding

2026-06-25

AI Summary

Existing temporal knowledge graph embedding methods struggle to simultaneously model the diverse mapping properties of relations—such as one-to-many and many-to-one—and temporal dynamics effectively. This work proposes a decoupled temporal rotation mechanism that applies time-aware rotational transformations to head and tail entities separately in complex space. To explicitly capture complex relational patterns, the approach introduces a relation-centered circular constraint, which restricts the rotated and translated head entity within a circular region centered on the relation. Evaluated on four standard temporal knowledge graph benchmarks, the proposed method achieves or surpasses state-of-the-art performance, demonstrating significantly enhanced capability in jointly modeling temporal evolution and relational diversity.

Abstract

In recent years, with the emergence of Temporal Knowledge Graphs (TKGs), research on learning entity and relation representations in TKGs has attracted increasing attention, giving rise to a large number of TKG embedding methods. TeRo is a simple and efficient temporal knowledge graph embedding approach. However, TeRo does not do well in modeling the mapping properties of various relations, such as one-to-many, many-to-one, and many-to-many. Meanwhile, it also has limitations in the expression of temporal information. To address these issues, we propose a novel TKG embedding method named TeRoR. This method divides the temporal evolution of entity embeddings, and conducts independent rotation transformations on head and tail entities in the complex vector space to strengthen temporal information modeling capacity. In terms of relational characteristics, we train a radius to constrain the rotated and translated head entities within a circular region centered on the tail entity, which effectively captures the diverse mapping properties of relations. Experimental results demonstrate that TeRoR achieves competitive performance against state-of-the-art models on four distinct TKG datasets.

Problem

• Temporal Knowledge Graph
• Relation Mapping Properties
• Temporal Information Modeling
• Knowledge Graph Embedding

Innovation

• Temporal Knowledge Graph Embedding
• Decoupled Temporal Rotation
• Relational Circular Region
• Complex Vector Space
• Mapping Properties