School of Interactive Computing Assistant Professor Alex Endert received a CAREER Award from the National Science Foundation for a project titled CAREER: Visual Analytics by Demonstration for Interactive Data Analysis.
The award, which will total $493,000 paid out over the course of five years, begins on May 1 and builds on Endert’s prior work on demonstration-based user interaction to create tools that make data science more usable and accessible to people without formal data science training.
The research will build knowledge in how people can visually demonstrate their questions about data. In turn, visual analytic system interfaces will need to change to interpret these demonstrations and perform the appropriate analytic operations. Finally, people will be able to leverage complex and powerful analytic functions without the need to provide formal parameterizations of the model being used.
“In today’s data-driven era, everyday decisions are becoming increasingly data-driven problems,” Endert explained. “While this provides opportunity for people to make better decisions, it requires technology for visual data analysis to become easier to use for people without formal data science training.”
It’s not just people in business who are constantly utilizing data to inform decisions. Everyday people encounter data on a daily basis – comparing car models, searching for houses, and more. Endert noted impactful areas of interest like health care and national security.
“This research will create new methods for people to interact with data, focusing on domains of interest to society including health care and national security,” said Endert, who added that he and his students will develop visual analytic prototypes released on the web, toolkits for developers to leverage, adopt and expand research, and provide empirical evidence to support the increase in usability.
A key challenge, Endert said, is fostering the interactive feedback loop between people and systems. The overall goal is to simplify aspects of this iterative process by building by-demonstration alternatives to existing control panels providing precise, yet complex controls.
“If successful, the proposed work has the potential to transform user interfaces for data science systems,” he said.
Endert is a part of a separate team of researchers that was recently awarded $2.7 million from the Defense Advanced Research Projects Agency Data-Driven Discovery of Models program to study similar advances in the accessibility of data science.