Graphlet-orbit Transitions (GoT): A fingerprint for temporal network comparison.

Area of Science:

• Network Science
• Complex Systems
• Data Mining

Background:

• Comparing temporal networks across diverse domains and scales presents significant challenges.
• Identifying characteristic and meaningful evolutionary patterns in dynamic network structures is crucial for understanding system behavior.

Purpose of the Study:

• To develop a novel method for comparing temporal networks by capturing their dynamic topological properties.
• To introduce a robust technique for identifying characteristic evolutionary patterns within temporal network data.

Main Methods:

• Introduced Graphlet-orbit Transitions (GoT), a temporal and topological network fingerprint.
• Extended graphlet concepts and utilized node orbits to define roles within subgraphs.
• Constructed a transition matrix to track node evolution through orbit trajectories.
• Developed the Orbit Transition Analysis (OTA) metric for comparing temporal networks based on these matrices.

Main Results:

• GoT provides rich and interpretable network characterizations.
• Experimental results demonstrate that GoT accurately groups synthetic networks based on graph models, exceeding state-of-the-art methods by over 30%.
• Analysis of real-world networks using GoT yields highly interpretable results, revealing characteristic orbit transitions.

Conclusions:

• GoT offers a powerful new approach for analyzing and comparing temporal networks.
• The method effectively captures dynamic network evolution and identifies meaningful patterns.
• GoT enhances our understanding of complex systems by providing interpretable insights into network dynamics.