Understanding and Supporting Users in Visual Network Exploration

Abstract

Network visualization tools are used in numerous domains to explore data. Various network visualizations have been designed to explore data from different perspectives. However, few resources investigate how analysts create and interact with the visualizations to explore networks in the wild. By analysts, we denote users varying in their background and level of expertise. Being in the wild signifies that analysts work outside controlled settings where no pre-designed tasks are set. This thesis focuses on studying how to understand and support analysts in the process of network visual exploration. In such a process, analysts need to learn about the concepts, the tool(s), the processes, the visualization(s) and interactions, and the workflow. They also need to ensure that they can apply what they have learned accumulatively on their data toward their goal(s). Thus, to understand and support analysts, we have applied and collected data through mixed methods: interaction logging, mini-questionnaire, and visualization's state annotation, running an intensive 6-week course, designing an analytical dashboard, and implementing a coaching program. We ran our studies using the Vistorian a web-based tool that offers 4 types of interactive network visualizations. This multimethod research led to the following contributions. Interaction logging allowed identifying 4 types of users based on their tool usage and advancement in the visual exploration process: demo users, data strugglers, single-session and multi-session explorers. To examine user types further, we designed a utility to capture and annotate visualizations' states, which we call bookmarks. We identified eight barriers that might cause analysts to struggle through the visual exploration process. We designed an analytical dashboard by specifying Key Performance Indicators (KPIs) and analyzing interaction logs accordingly. Those KPIs informed the assessment of the tool, the visualizations, the help resources, and the users' exploration. To support analysts, we designed a self-regulated guide for network visual exploration, which we call a roadmap. The roadmap describes step-by-step the processes of network visual exploration and associated activities. We also described 16 exploration strategies analysts follow, classified into three categories. We found 4 distinguished analyst groups based on how they aim to explore whether with/without research goals and/or data, which we call roadmap pathways. We evaluated the roadmap through a coaching program and found that it plays a crucial role in teaching networks visual exploration. Those findings have implications for the network's visual exploration through enhancing the design of its tools and associated educational efforts, mitigating barriers within, and supporting various user types.