Spatiotemporal Interactive Modeling of Event-based Dynamic Networks.

Area of Science:

• Network Science
• Data Science
• Computational Social Science

Background:

• Event-based dynamic networks are prevalent in traffic, biology, and social systems.
• Modeling these networks requires understanding how events influence subsequent occurrences based on spatial and semantic relationships.
• Existing models often struggle to fully integrate spatial structure and historical event data.

Purpose of the Study:

• To propose a novel model, the spatiotemporal interactive Hawkes process (SIHP), for analyzing event-based dynamic networks.
• To explicitly model the rate of interaction events between network nodes by incorporating historical data and spatial/semantic proximity.
• To enhance the understanding of network dynamics and influence patterns.

Main Methods:

• Developed the spatiotemporal interactive Hawkes process (SIHP) model.
• Incorporated spatial structure knowledge as a graph and applied graph regularization.
• Utilized an alternating direction method of multipliers (ADMM) framework for model parameter estimation.

Main Results:

• The SIHP model effectively learns influence patterns from historical events.
• The model successfully integrates spatial structure knowledge to understand network dynamics.
• Validated through numerical experiments and a real-world case study on New York yellow taxi data.

Conclusions:

• The proposed SIHP model provides a robust framework for analyzing event-based dynamic networks.
• The method accurately captures spatiotemporal dependencies and influence propagation.
• Demonstrates significant effectiveness in real-world applications, such as traffic pattern analysis.