Visual analysis of complex social networks
Zeqian Shen · 2009
A social network represents relationships between actors. Analyzing a social network can provide structural intuition according to the ties linking actors and lead to improved understanding of social processes. Visualization systems have helped analysts to create a drawing of a social network easily and investigate the visualization interactively. Lately, the emerging Internet provided opportunities to collect social networks on a scale far larger than previously possible. Those traditional visualization techniques that were designed for networks of tens or hundreds of nodes are no longer effective. In addition to the large scale, the networks show much more structural and semantic complexities. On the other hand, social network analysis is becoming an active area of study beyond sociology, because of the tremendous value of its findings. More and more researchers from other fields without expert knowledge in sociology are trying to understand social networks from their own perspectives. Therefore, intuitive visual analytics tools that are equipped with advanced visualization techniques and support exploratory analysis processes for average users are desired. This dissertation addresses these challenges arising in visual analysis of complex social networks. I present a family of visual analytics tools that enable interactive exploration and discovery for complex social network analysis. Advanced structural and semantic abstraction techniques have been invented to effectively reduce visual complexity during the analysis. Compact but insightful visual representations have been designed to convey the additional semantic information (e.g., temporal and spatial information) of the complex networks. I have also developed illustrative techniques to close the gap between traditional graph visualization techniques and advanced analytic tasks. A flexible and extensible force-directed label placement method tailored for graphs is introduced. A design of augmenting adjacency matrices with interactive path visualization to remove the intrinsic limitation of a matrix representation is also presented. Case studies and evaluations have confirmed the practical values of these tools on analyzing P2P community sharing networks, collaborative movie networks, terrorism networks, mobile social networking data, etc.