Mahyar, Sarvghad, Jasim Win Best Paper at EuroVis2022 for Critical Inquiry Into Civic Text Visualization

Assistant Professor Narges Mahyar, Research Assistant Professor Ali Sarvghad, and doctoral student Mahmood Jasim of the Manning College of Information and Computer Sciences (CICS) at UMass Amherst, along with lead author Eric Baumer of Lehigh University, received the Best Paper Award at EuroVis 2022 (Eurographics Conference on Visualization) in Rome, Italy on June 17 for “Of Course it's Political! A Critical Inquiry into Underemphasized Dimensions in Civic Text Visualization.” The paper provides conceptual tools to help mitigate potential issues when researchers, designers, and practitioners apply text visualization techniques to make sense of public input in civic decision-making.

Often, public input is collected by governments or civic organizations as free-form texts representing opinions, suggestions, or requests pertaining to the provision of services. To help identify priorities and uncover hidden insights from public input, decision-makers often use text visualization tools to make sense of these massive volumes of text. As the authors explain, these tools often focus only on the analytic aspect and not the potential political aspects of text visualization.

“The political aspect of visualization becomes crucial in civic decision-making,” explains Mahyar. “While text visualization tools are powerful, their current focus on being purely analytic has the effect of drawing attention away from the political aspects of public input.”

This emphasis on treating civic text visualization as purely analytic raises several issues, the authors explain, including the marginalization of minoritized voices and the suppression of uncertainty and potential infiltration of biases. To address these issues, they offer a series of conceptual dimensions that discuss potential shifts between analytic and political orientations.

Utilizing their conceptual dimensions could lead to more inclusive and democratic civic text visualizations. For example, rather than designing for a single analyst with a single point of view, visualizations could be designed around multiple relationships among different stakeholders representing different communities. Other dimensions suggest design approaches that emphasize interpretation, rather than prescription of meaning, politically aware approaches to include paradata to highlight where the data comes from, who is included and excluded, what data is perceptible and what is elided. Finally, design approaches that balance complexity while paying attention to who can or can not use the system. 

“We believe these dimensions can act as conceptual tools to help visualization designers and practitioners to make design decisions that enhance democratic decision-making,” says Mahyar. “In fact, we suggest that these conceptual dimensions can help attune to political aspects of visualizations in other domains, not just within politics.”