Northwestern University

Thu 2:00 PM

CS + X: Journalism Colloquium - Prof. Jessica Hullman, Assistant Professor in the Information School at University of Washington, "Improving Data Reasoning Using Visualization and Automation"

Prof. Jessica Hullman

When: Thursday, March 15, 2018
2:00 PM - 3:00 PM  

Where: Technological Institute, Room L440, 2145 Sheridan Road, Evanston, IL 60208 map it

Audience: Faculty/Staff - Student - Public - Post Docs/Docs - Graduate Students

Contact: Brianna Mello   847.467.6558

Group: Electrical Engineering & Computer Science

Category: Lectures & Meetings


The EECS Department welcomes Prof. Jessica Hullman, Assistant Professor in the Information School at University of Washington.

Hullman will present a talk entitled "Improving Data Reasoning Using Visualization and Automation", on Thursday, March 15 at 2:00 PM in the Tech Room L440.

Abstract: Many non-scientists interact with data and scientific information through online media outlets. Good graphical representations are critical for these applications. Unfortunately, the professional expertise required for constructing engaging, understandable summaries does not scale. Automation may help, but purely computational solutions cannot match experts' careful reasoning. Research in my lab embeds principles learned from expert journalists and designers in tools to improve data cognition. Our goal is to create systems to that make it easier to author and interpret comprehensible data representations. I'll detail two major research efforts that demonstrate this approach.

I'll first describe tools that address the specific strategy of using analogies to help readers comprehend measurements, a task that currently requires time and expertise to do well. I'll describe tools that my group developed by identifying criteria for effective analogies, devising data extraction pipelines that make use of online measurement data, semantic networks, and crowdsourcing, and developing automated algorithms that enact common analogy strategies like reunitization and proportional analogy at scale.

The second line of research addresses the broader task of building stories through multiple-view visualizations. Multiple views are ubiquitous in analysis and communication, but the state-of-the art in visualization authoring systems, including Tableau, embed principles designed to guide authors toward the best single visualizations. Using these tools can easily result in sets of visualizations (dashboards, data reporting, sequential presentations) that contradict our expectations for analogous use of encodings for data across views. I'll describe what we've learned from studying how visualization authors trade-off multiple view and single view concerns, and how these observations are leading us to re-envision the visualization authoring paradigm.

Bio: Jessica Hullman is an Assistant Professor in the Information School and Adjunct Assistant Professor in Computer Science & Engineering at the University of Washington. She co-directs the Interactive Data Lab and is a member of the cross-departmental HCI group DUB. The goal of Dr. Hullman's research is to create computational tools that can scale effective data graphics and summaries to more contexts. She is particularly inspired by how science and data are presented to non-expert audiences in data and science journalism, where a shift toward digital news provides opportunities for informing through interactivity and visualization. Successful visualizations in these contexts anticipate people's internal representations of complex concepts like uncertainty, measurement, or relational data, but take significant time and expertise to create. By identifying principles in experts' designs, instantiating these in tools, and studying the impacts on non-experts' reasoning, Dr. Hullman's work deepens understandings around data cognition while addressing current challenges in data-driven communication. Her specific research topics include visualizing uncertainty, design and evaluation of multiple view visualizations, automated visualization generation for news, interaction with prior knowledge in visualizations, text simplification for summarizing scientific findings, and the role of rhetoric in visualization use. Dr. Hullman has co-authored over 30 peer-reviewed papers and received paper awards from ACM's CHI conference and IEEE's InfoVis conference. Her work has been featured in journalism and data storytelling outlets like Data-Driven Journalism and the OpenVis and Tapestry conferences, among others. She is the recipient of an NSF CRII Award and a Google Faculty Award, among other grants. Dr. Hullman completed her Ph.D. in Information Science at the University of Michigan in late 2013. She held an inaugural Tableau Software Postdoctoral Fellowship in Computer Science at the University of California Berkeley in 2014 prior to joining the University of Washington.

Hosted by: CS Division & Medill School of Journalism  

Add Event to Calendar

Add Event To My Group:

Please sign-in