Leveraging Evolution Dynamics to Generate Benchmark Complex Networks with Community Structures

Abstract

The past decade has seen tremendous growth in the field of Complex Social Networks. Several network generation models have been extensively studied to develop an understanding of how real world networks evolve over time. Two important applications of these models are to study the evolution dynamics and processes that shape a network, and to generate benchmark networks with known community structures. Research has been conducted in both these directions, relatively independent of the other. This creates a disjunct between real world networks and the networks generated as benchmarks to study community detection algorithms. In this paper, we propose to study both these application areas together. We introduce a network generation model which is based on evolution dynamics of real world networks and, it can generate networks with community structures that can be used as benchmark graphs. We study the behaviour of different community detection algorithms based on the proposed model and compare it with other models to generate benchmark graphs. Results suggest that the proposed model can generate networks which are not only structurally similar to real world networks but can be used to generate networks with varying community sizes and topologies.

References in corpus (10)

• Fast unfolding of communities in large networks
• Maps of random walks on complex networks reveal community structure
• Benchmark graphs for testing community detection algorithms
• Statistical Mechanics of Community Detection
• Benchmarks for testing community detection algorithms on directed and weighted graphs with overlapping communities
• Social Structure of Facebook Networks
• Community detection in networks: Structural communities versus ground truth
• A Large-Scale Community Structure Analysis In Facebook
• Empirical Comparison of Algorithms for Network Community Detection
• Growing community networks with local events