Abstract
Temporal knowledge graphs (TKGs) are dynamic heterogeneous multirelational graphs whose topological structures evolve over time. Reasoning over TKGs involves forecasting future events based on complex historical structural information. Efficiently modeling TKGs' structural interactions at each timestamp and the evolution timelines is crucial for later reasoning. However, previous approaches simply learned shallow correlations over TKGs but ignored the heterogeneous deep hierarchy for reasoning. Furthermore, previous evolution modeling techniques for TKGs merely capture the general trends of a given chronological sequence, forgetting the fine-grained event distributions in history. To this end, we propose $\bf {H^{2}}$-$\bf {Evo}$, which learns deep $\bf{\underline{H}}$ierarchy of intra-time $\bf{\underline{H}}$eterogeneous interactions and hyperbolic $\bf{\underline{Evo}}$lution of inter-time hierarchical dependencies. Specifically, to learn the heterogeneous deep hierarchy in TKGs, we design a heterogeneous hyperbolic message-passing framework with inter-hierarchy perception, intra-hierarchy correlation, and multilayer propagation. It effectively captures the heterogeneous and hierarchical structures in TKGs. To properly capture the temporal evolution of hierarchical structures in TKGs, we devise a novel hyperbolic dual-attentive mechanism, which jointly learns the fine-grained time-varying evolution patterns of TKGs' hierarchical structure and its period-varying evolving trends for later reasoning. Extensive experiments on four public benchmarks demonstrate that $\bf {H^{2}}$-$\bf {Evo}$ has achieved state-of-the-art reasoning performance.
Funding Information
  • National Key Research and Development Program of China (2023YFF0905503)
  • National Natural Science Foundation of China (62472188)
  • The Education University of Hong Kong Start-up Funding (R4400)
  • EdUHK Seed Funding (R4539)

This publication has 51 references indexed in Scilit: