In the field of information visualization, researchers and developers have created many types of visualizations, or visual depictions of information. User interface designers often coordinate multiple visualizations, taking advantage of the strengths of each, to enable users to rapidly explore complex information. However, the combination of visualizations and coordinations needed in any given situation depends heavily on the data, tasks, and users. Consequently, the number of needed combinations explodes, and implementation becomes intractable.
Snap-Together Visualization (Snap) is a conceptual model, user interface, software architecture, and implemented system that enables users to rapidly and dynamically construct coordinated-visualization interfaces, customized for their data, without programming. Users load data into desired visualizations, then create coordinations between them, such as brushing and linking, overview and detail, and drill down.
This dissertation presents four primary contributions. First, Snap formalizes a conceptual model of visualization coordination that is based on the relational data model. Visualizations display relations, and coordinations tightly couple user interaction across relational joins.
Second, Snap's user interface enables the construction of coordinated-visualization interfaces without programming. Data users can dynamically mix and match visualizations and coordinations while exploring. Data disseminators can distribute appropriate interfaces with their data. Interface designers can rapidly prototype many alternatives.
Third, Snap's software architecture enables flexibility in data, visualizations, and coordinations. Visualization developers can easily snap-enable their independent visualizations using a simple API, allowing users to coordinate them with many other visualizations.
Fourth, empirical studies of Snap reveal benefits, cognitive issues, and usability concerns. Six data-savvy users successfully, enthusiastically, and rapidly designed powerful coordinated-visualization interfaces of their own. In a study with 18 subjects, an overview-and-detail coordination reliably improved user performance by 30-80% over detail-only and uncoordinated interfaces for most tasks.
Snap has proven useful in a variety of domains, including census statistics and geography, digital photo libraries, case-law documents, web-site logs, and traffic incident data.
Some individual portions: