Temporal knowledge graph model improves fact prediction accuracy
Biquaternionic Space with Complex-valued Attention for Temporal Knowledge Graph Completion
Machine LearningArtificial Intelligence
Summary
Knowledge graphs are like big maps of facts that can change over time, and it's hard to guess missing facts accurately. The authors made a new model called BSCA that uses math from biquaternionic spaces to better understand how facts change over time. Their model can adjust how much it pays attention to different parts of the information, depending on time and type of relationships. When tested on several datasets, BSCA was better at guessing missing facts than earlier methods, especially on one called GDELT.
What this means in practice
- •For data engineers: Improve the accuracy of systems that predict missing time-based relationships in knowledge graphs for better data completion.
- •For knowledge graph developers: Create dynamic entity representations in temporal knowledge graphs that adapt to changes in time and relation contexts to enhance inference.
Authors
Rushan Geng, Cuicui Luo
Abstract
Temporal knowledge graph embedding (TKGE) models infer missing facts in knowledge graphs that evolve over time. Many existing models use a single geometric space, which can limit their ability to represent diverse relational patterns, or treat entity representations as static. We propose Biquaternionic Space with Complex-valued Attention (BSCA), a TKGE model that combines circular and hyperbolic rotations within a unified biquaternionic framework. A complex-valued attention mechanism adaptively fuses time-conditioned and relation-conditioned entity representations, allowing them to vary with temporal and relational context. Experiments on five benchmark datasets show competitive performance across datasets, with the largest improvement on GDELT: BSCA achieves an MRR of 52.1\%, compared with 38.1\% for the strongest baseline in our comparison.