Memory in network flows and its effects on spreading dynamics and community detection

Area of Science:

• Network science
• Complex systems analysis
• Computational social science

Background:

• Random walks on networks are standard for modeling spreading processes.
• First-order Markov models, commonly used, ignore directional dependencies in flow.
• This limitation impacts community detection, ranking, and spreading analysis.

Purpose of the Study:

• To investigate the impact of second-order Markov dynamics on network analysis.
• To explore how accounting for memory in network flows enhances understanding of system organization.

Main Methods:

• Analysis of pathways using second-order Markov models.
• Application to real-world datasets, including air traffic and scientific communication networks.

Main Results:

• While disease spreading showed marginal effects, community detection, ranking, and information spreading were significantly impacted.
• Second-order models revealed actual travel patterns in air traffic.
• Multidisciplinary journals were uncovered in scientific communication networks.

Conclusions:

• Ignoring second-order Markov dynamics has significant consequences for network analysis.
• Accounting for higher-order memory in network flows improves the understanding of real-world system organization.
• This approach enhances community detection, ranking, and information spreading analysis without additional assumptions.