CAREER: Context-Aware Visual Analytics Systems: Evolving the One-Size-Fits-All Approach to Design and Evaluation

Washington University, Saint Louis MO

Investigators

Abstract

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2). Large-scale data analysis is increasingly vital in domains from national security to e-commerce. Consequently, we have seen an increasing number of analysis and visualization tools to provide automated support for human analysts. However, the current tools tend to offer support with a “one size fits all” approach that fails to consider differences in analysts’ cognitive skills and styles, affecting both how they do their jobs and how well they do them. This project envisions visual analytics systems that sense the user and their goals to actively support their analysis by providing the right information at the right time. The work will produce: (1) a theoretical understanding of how individual differences impact analytic workflows and which traits are most predictive; (2) algorithms for analyzing a user’s interactions with visualizations to infer their cognitive profile and analysis goals; and (3) techniques and theory for using those inferences to give more personalized suggestions. These three aims help facilitate a human-technology partnership where a better understanding of the analyst enables a more practical machine partner, leading to greater exploration of the data space and more efficient hypothesis generation and decision-making. This project systematically investigates the impact of individual differences on a broad range of tasks and visualization designs. The general approach involves leveraging low-level behavioral data such as mouse interactions to model the user’s cognitive profile, attention, and workflow. These data and models will allow the researchers to assess and predict cognitive traits that affect analysts’ visualization strategies and effectiveness. Based on these models, the research team will develop context-aware visual analytics tools that offer personalized guidance and suggestions based on the learned user characteristics and goals, calibrating the type and timing of assistance provided based on a given user profile and workflow. Altogether, this research agenda incorporates situational context into the design and evaluation pipelines to create broadly usable visualization tools. It develops new theoretical foundations, algorithms, and techniques that would result in a more symbiotic relationship between the human and the visual analytic tool during data analysis. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.