Community detection using efficient modularity optimization method: labelmod with single and multi-layer graphs
Graph clustering is a field of study that helps reveal characteristics of communities. Systems can be viewed as networks and form communities in various areas such as biology, computer science, engineering, economics, and politics. A clustering algorithm is a tool that detects communities and it can be also considered as a pre-processing step to study the characteristics of detected communities. Many efforts were made to develop a well performing clustering algorithm in different types of networks. In recent literature, a concept of multi-layer graphs emerged, and clustering algorithms are being developed to detect communities in the multi-layer graphs. In this thesis, we propose a clustering algorithm that can be applied to both single-layer and multi-layer graphs. We test the algorithm on simulated data and real data in both single-layer and multi-layer graphs. Four performance measures were used to evaluate the performance of the proposed algorithm. We also study how the performance measures are correlated with each other and what the effects of parameter, presented in the proposed algorithm are. The thesis concludes with summary of research findings and directions of the future research.
Lee, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our