Ego-networks, networks that emphasize relationships between a specific node (called the ego) and its neighbor nodes (called the alters), have been extensively studied in the area of network analysis. Generally, ego-networks are dynamic in nature and their topological structures and properties change over time. Exploring the alterations of ego-networks can provide profound insight and better understanding of how these networks evolve. However, structural changes and attribute changes are difficult to present in traditional node-link diagrams. Most existing dynamic ego-network visualization techniques only focus on structural changes of a single ego-network, while ignoring the changes of other network attributes. In addition, comparisons of multiple ego-networks are not well supported in...[ Read more ]

Ego-networks, networks that emphasize relationships between a specific node (called the ego) and its neighbor nodes (called the alters), have been extensively studied in the area of network analysis. Generally, ego-networks are dynamic in nature and their topological structures and properties change over time. Exploring the alterations of ego-networks can provide profound insight and better understanding of how these networks evolve. However, structural changes and attribute changes are difficult to present in traditional node-link diagrams. Most existing dynamic ego-network visualization techniques only focus on structural changes of a single ego-network, while ignoring the changes of other network attributes. In addition, comparisons of multiple ego-networks are not well supported in previous works. In this thesis, we discuss the major challenges in dynamic ego-network visualization, starting from how to encode temporal information to reveal structural changes of an ego-network. Then, we develop a visual analytic system to reveal attribute changes and evolutionary trends of an ego-network, with a case study from a real world dataset. Finally, we propose a new method to visualize ego-network similarity, which allows analysts to better compare multiple ego-network properties.