Jessica Ruth Hullman

Research - Selected projects

Narrative visualization

I have worked in the area of narrative visualization, or the use of statistical graphics to tell stories around data. I've demonstrated approaches for automatically generating and annotating visualizations to accompany news, similar to the way professional designers annotate visualizations to draw attention to important patterns (most recently applied to news maps). I proposed visualization rhetoric as a framework for understanding persuasion through visualizations. I've studied how contextual factors like viewing order affect interpretations and proposed a graph-based algorithm for designing "data stories".

Supporting Data Cognition Among Broader Audiences

Many visualization systems focus on expert users, such as trained statisticians or researchers. I am interested in understanding how visualizations and data tools can be used effectively by non-expert audiences as well as trained analysts. Inspired by the “concrete scales” created by graphic designers to make measurements more relatable), I am developing measurement databases and automated algorithms to perform measurement re-expressions as a user reads a text article (e.g., 380m ft is about twice the height of the Space Needle). Recent work produces personalized analogies of spatial measurements like distances and areas.



Many people, including analysts, find uncertainty and probability difficult to reason about when working with data. I’m developing a method for visualizing uncertainty more concretely as a set of possible outcomes. A set of hypothetical data samples is generated and presented in an animated or interactive format. By watching possible outcomes "play out", the user gains a better sense of which data patterns are reliable and which are not. The approach generalizes to a number of data inputs and complex visualization types that lack uncertainty representations, like choropleth maps and network diagrams, and lead to more accurate understanding of the reliability of observed relationships in multivariate data than standard approaches like error bars.



Hypothetical samples are also beneficial when presenting visualizations to an online crowd to evaluate a hypothesis. For example, mechanisms that present each crowd member with several pieces of visual evidence and ask them to make an overall judgment are affected by order biases, while presenting a single visualization to a crowd can result in a systematic bias across the crowd.