Heterogeneous, Temporal, and Multi‑layer Networks: From Dynamic Graphs to Event‑Driven Knowledge Systems

Real-world systems, such as social platforms, biological networks, transportation networks, and financial markets, feature complex interactions, continuous evolution, and interconnected layers. Traditional network models rely on static, single-layer assumptions. As a result, they struggle to capture diverse entity types, time-varying interactions, and cross-layer dependencies.Recent advances in dynamic graph learning, multi-layer network modeling, knowledge graph reasoning, and event-driven architectures offer promising tools for complex systems. However, these fields have developed in silos. Major challenges remain in designing unified frameworks that simultaneously handle heterogeneity, temporal dynamics, and multi-layer structures. Uniting these research areas is essential to build intelligent systems capable of reasoning over streaming, multi-source, and context-rich network data.This Research Topic focuses on developing unified frameworks for modeling and reasoning over complex networks that exhibit heterogeneous structures, temporal evolution, and multi-layer dependencies. Existing methods often isolate these dimensions. They treat temporal dynamics as external factors, multi-layer interactions as detached views, or node and edge heterogeneity through disconnected preprocessing steps. These fragmented approaches cause information loss, flatten rich semantics, and fail to adapt to live environments.We aim to advance theories, algorithms, and computational architectures that combine heterogeneous, temporal, and multi-layer dimensions into a single network paradigm. Key directions include robust graph representation learning, continuous-time and event-driven modeling, semantic knowledge reasoning, and scalable pipelines for massive network streams. We also welcome realistic benchmarks and reproducible evaluation protocols to accelerate real-world deployment. By connecting graph learning, network science, and knowledge systems, this collection seeks to deliver adaptive, interpretable, and scalable methods for network analysis.We invite original research and review articles addressing, but not limited to, the following themes: - Temporal graph neural networks for heterogeneous and multiplex structures - Event-driven dynamic knowledge graph completion and reasoning - Multi-layer community detection and longitudinal evolution tracking - Streaming algorithms and online learning for time-varying multiplex networks - Representation learning for temporal hypergraphs and higher-order network motifs - Causal inference and counterfactual reasoning on event-based graph data - Scalable systems, storage architectures, and indexing for temporal multi-layer graphs - Domain applications across social networks, epidemiology, algorithmic finance, IoT, and smart infrastructure - Theoretical foundations of temporal reachability, dynamic centrality, and resilience in multi-layer topologies - Neuro-symbolic integration combining symbolic domain rules with neural models for explainable temporal predictions Topic editors: Pengwei Wang, Hongda Qi Source: https://www.frontiersin.org/research-topics/85944/heterogeneous-temporal-and-multi-layer-networks-from-dynamic-graphs-to-event-driven-knowledge-systems