HOMEPAGE
       
Research
A web application to estimate and visualize sentiment—an attitude or thought based on emotion—for tweets from Twitter. Each tweet is shown as a circle in an emotional scatterplot, with pleasure on the horizontal axis and arousal on the vertical axis. Other methods of viewing the tweets include topic clusters, a heatmap of tweet counts by sentiment, tag clouds of frequent terms, a timeline of when tweets were posted, and a map of where tweets were posted

We have recently updated our web application for visualizing tweet sentiment. Recent tweets are requested based on user-chosen keywords. An estimate of each tweet's sentiment is made from its text. The results are visualized in numerous ways: in an emotional scatterplot with horizontal pleasure and vertical arousal axes, clustered by topic, in a heatmap that counts tweets by sentiment bins, as a tag cloud of frequent terms, as a timeline of when tweets were posted, as a map of where tweets were posted, and as a list of tweets and their associated sentiment measurements.

Projects
Perceptual visualization of flow in a simulated supernova collapse
Experimental studies of low-level human visual perception. Our goal is a set of fundamental perceptual guidelines for visualization design.
Resources
Multivariate visualization of memory chip validation data
A study of multivariate data, where each data element encodes values for multiple data attributes. Our goal is a set of guidelines for designing effective multivariate visualizations.
Resources
Nonphotorealistic visualization of weather conditions over South America
Exploring ways to engage the human visual system through the use of aesthetic visual cues. Our goal is a method to improve memory for detail at important locations in a visualization.
Resources
Visualizing the sentiment of tweets containing the keyword "Boston"
An investigation of sentiment—the implied emotion—in short text snippets like tweets, RSS feeds, or Facebook wall posts. Our goal is a set of methods to estimate and visualize sentiment in text.
Resources
  • Tweet Viz, a web app to query and visualize tweets from Twitter
Visualizing climatology data over the western United States
Integration of perceptual guidelines into a planning system that uses probabilistic utility to suggest visualizations for a target dataset and analysis tasks. Our goal is a tool that collaborates with an analyst to generate a set of perceptually optimal visualizations that fit the analyst's needs.
Resources
Visualizing an ensemble from the Relativistic Heavy Ion Collider
A study of ensembles: large collections of related datasets, for example, from physics or climatology simulations. Our goal is a scalable approach to to manage and visualize ensemble data.
Resources
Visualizing association rules for climatology data
Integration of applied mathematical techniques like rule mining, clustering, aggregation, summarization, and outlier detection to structure a dataset prior to visualization. Our goal is to intelligently summarize large datasets by highlighting features of interest at multiple levels of detail.
Resources
2012 Texas election results by Congressional district
Visualizations of Presidential, Senate, House, and state Governor election results in each state and Congressional District in the U.S. Our goal is a design that shows how common groups of individuals voted for different elected offices.
Resources
Visualizing plankton densities in the northern Pacific Ocean
An investigation of estimating open-ocean plankton densities based on sea surface conditions. Our goal is to use the improved methods to better predict sockeye salmon migration patterns.
Resources
Nonphotorealistic visualization of flow in a simulated supernova
An application of perceptual visualization techniques for exploring slices through a simulated supernova collapse. Our goal is a multivariate method that highlights flow properties like vortices, paths, and shock fronts.
Resources
Visualizing the 2013 Baltimore–San Francisco Super Bowl game
An investigation of parsing and visualizing play-by-play text for NFL games. Our goal is a design that is simple, effective, and readily accessible to the general public.
Resources
Research

Perception

Multivariate

Engagement

Text

Navigation

Mixed Initiative

Ensemble

Data Analytics
Applications

Election

Marine Biology

Astrophysics

Football

I am investigating visualization techniques that support rapid, accurate, and effective exploration and analysis of large, complex, multivariate datasets. Many of our approaches harness visual perception. This allows much of the analysis to be performed automatically by the low-level visual system. Research in an area of cognitive psychology known as preattentive processing forms the foundation for our perceptual guidelines. A detailed overview of preattentive processing is available for those who are interested.

Publications

A full list of publications is available, with links to citation information, abstracts, and PDF for most of the papers. A PDF version of my CV is also available. Some of our recent publications include the following.

Schedule

If you want to schedule a meeting with me, please check my calendar for availability, then email me so I can confirm the time is still open.

Students

PhD.   Lihua Hao, Kalpesh Padia.
MS.   RaviTeja Manda, Pongobinath Sivashanmugam.

A complete list of past students is also available.

Courses

CSC 561, Introduction to Graphics
MWF 9:35-10:25  1229 EB-II

CSC 600, Graduate Orientation
F 2:20-3:10  1025 EB-II

CSC 463/563, Graphics & HCI for Mobile Apps
MWF 9:35-10:25  1005 EB-I

CSC 541, File Structures
MWF 10:40-11:30  1007 EB-I