Grant: Discovery Grants Program - Individual

Visual analytics systems provide complex visual workspaces (CVWs) where users manipulate datasets to explore hypotheses, find insights, and make decisions. CVW systems are now common in many areas, such as genomics, logistics, and data analytics. As users gain experience with their data, they should be able to work faster and more accurately - but current CVW systems limit the performance levels that users can attain. The first problem is that current systems do not enable expertise for fundamental interactions such as navigating, revisiting previous locations, comparing elements, or managing views. Second, most CVWs do not provide expert shortcuts for interactive tasks, limiting a user's ability to improve performance. Most operations are based on direct manipulation, which is good for novices, but leaves experienced users "stuck in beginner mode." Third, CVW systems provide little support for users who want to improve. To support expertise development, systems need to help users expand their command vocabularies, help them switch to expert methods, and help them transfer just-learned expert techniques to real-world tasks. As expert capabilities are added to CVW systems, providing effective mechanisms to help users move towards expertise becomes critical. The overall goal of this research program is to substantially improve expert performance and expertise development in complex visual workspaces. As knowledge work increasingly involves working with large and complex datasets, it becomes increasingly important to provide powerful interaction techniques that have a higher performance ceiling for experienced users, as well as explicit support for users as they develop their expertise. Supporting expertise in CVWs will greatly improve user performance and user satisfaction for visual analytics, exploratory visualization, and decision making with complex data. The overall goal will be pursued through three key objectives. First, we will develop novel expert techniques for CVWs that capitalize on these systems' visual, spatial, and configuration complexity. We will build techniques based on spatial location memory, interaction history, and commands that can accelerate actions in direct-manipulation environments. Second, we will promote expertise development in CVWs. We will create a framework of vocabulary expansion and new vocabulary-presentation methods specific to CVWs; we will build new mechanisms for accelerating navigation expertise with visual guides; and we will build techniques that improve adoption of memory-based expert methods. Third, we will improve persistence and transfer of expertise to CVWs. We will compare retention over time for different types of expert methods, we will develop new techniques for improving persistence of memory-based techniques, and we will identify new mechanisms for ensuring that users actually make use of expert methods in real-world systems and tasks.