Multiscale Community Blockmodel for Network Exploration.

Area of Science:

• Network Science
• Sociology
• Computational Social Science

Background:

• Real-world networks display complex phenomena like hierarchy, multiscale interactions, and diverse community topologies.
• Existing network analysis methods often fail to capture the interplay of these phenomena.

Purpose of the Study:

• To propose a nonparametric Multiscale Community Blockmodel (MSCB) for modeling hierarchical social communities.
• To address the selective membership of actors and network generation from within- and cross-community interactions.

Main Methods:

• Utilized the nested Chinese Restaurant Process for automatic inference of hierarchy structure.
• Developed a collapsed Gibbs sampling algorithm for posterior inference.
• Validated the model using synthetic networks.

Main Results:

• The MSCB model effectively captures hierarchical organization and community structures.
• Demonstrated the model's capability in analyzing real-world datasets.
• Successfully inferred network properties from selective actor memberships and interactions.

Conclusions:

• The MSCB provides a robust framework for analyzing complex network structures.
• The model offers insights into the generation of hierarchical and multiscale networks.
• Applicable to diverse real-world networks, including ecological and citation networks.