Detecting sequences of system states in temporal networks

Area of Science:

• Complex Systems Science
• Network Science
• Data Science

Background:

• Time-evolving systems generate interaction data forming temporal networks.
• These networks reflect underlying system states, crucial for understanding system dynamics.
• Existing methods may lack the granularity to infer discrete states from complex temporal interactions.

Purpose of the Study:

• To develop a coarse-grained method for assigning discrete states to time-evolving systems.
• To infer the sequence of these states from interaction data within temporal networks.
• To demonstrate the method's applicability across diverse scientific domains.

Main Methods:

• Combines a graph distance measure with hierarchical clustering.
• Applies the method to analyze social temporal networks.
• Infers discrete states representing system activities and temporal patterns.

Main Results:

• Successfully inferred distinct activities in a school setting.
• Differentiated between weekday and weekend states using social network data.
• Validated the method's capability in identifying system states from empirical data.

Conclusions:

• The proposed method offers a robust approach for coarse-grained state inference in temporal networks.
• Applicable to diverse fields including neuroscience, organizational dynamics, and ecology.
• Provides a framework for understanding dynamic processes in complex systems.