Networks in time and space : visual analytics of dynamic network representations

Abstract

Networks are abstract and flexible data structures that model entities and relationships between them. Network visualization encompasses techniques, designed to produce aesthetic and scalable layouts of such data. In real-world applications data changes over time and investigating and analyzing these dynamics is of interest, motivating the study of dynamic network visualization. Dynamic network visualization explores how networks evolve and how these changes can be effectively conveyed. The goal is to construct representations depicting both structural and temporal information in a readable manner. This is a challenging task and current approaches often optimize standard network representations (i.e., node-link diagrams) to simultaneously portray topology and dynamics using animation. Alternative visualization modalities are seldom explored, presenting an interesting research direction. As network connectivity can be cumbersome to visualize, approaches make use of space to improve readability and implement aesthetic criteria, often presenting scalability concerns. These limitations pose interesting research questions that are investigated in this dissertation. Namely, how can we augment standard network visualization techniques to convey dynamics, which techniques are appropriate, effective, and avoid information overload, and are alternative visualization metaphors for dynamic networks useful? We contribute to dynamic network visualization by formally evaluating the design space and investigating novel dynamic network visualization approaches. Our results show promise for alternative visualization modalities and our evaluation of structural and temporal encodings highlights the usefulness of under-investigated combinations in related literature. Furthermore, we outline research challenges that present future work opportunities and provide recommendations to support researchers in developing dynamic network visualization approaches.