Information diffusion backbones in temporal networks.

Area of Science:

• Complex Networks
• Information Diffusion Dynamics
• Network Science

Background:

• Understanding information diffusion in temporal networks is vital for optimizing spread.
• Identifying critical node pairs that drive diffusion is a key challenge.
• Existing models often overlook the specific role of node pair temporal connections.

Purpose of the Study:

• To determine which node pairs are most likely to participate in information diffusion trajectories.
• To investigate the relationship between node pair diffusion likelihood and local temporal connection features.
• To identify actionable insights for maximizing information spreading through targeted interventions.

Main Methods:

• Utilized the Susceptible-Infected (SI) model to simulate information diffusion on real-world temporal networks.
• Constructed an "information diffusion backbone" (G_B(β)) representing the likelihood of node pairs appearing in diffusion trajectories.
• Analyzed the relationship between backbones generated with varying infection probabilities (β).

Main Results:

• The backbone topology converges to G_B(β→0) (integrated weighted network) for low infection probabilities and G_B(β=1) for high probabilities.
• A specific local connection feature, encoding contact timing, effectively identifies high-weight node pairs in the G_B(β=1) backbone.
• This highlights the critical role of temporal contact information in diffusion.

Conclusions:

• Node pair importance in information diffusion is strongly linked to their local temporal connection features.
• Contact timing is a significant factor in determining a node pair's contribution to global information spread.
• The findings offer a framework for understanding and enhancing information diffusion in dynamic networks.