College Park, Md.—When Twitter traffic is laid out on a graph, revealing patterns emerge. Data from the night President Obama announced that Osama bin Laden was dead show a sharp drop in posts at the moment the news was revealed, as if the country took a collective gasp.
“You can see the various aspects of the country nervously watching,” said Jimmy Lin, an associate professor at the University of Maryland’s College of Information Studies, in a talk, “Visual Analysis at Twitter.” He spoke Wednesday at Maryland’s Human-Computer Interaction Lab Symposium.
Mr. Lin’s talk stemmed from observations he’s made during a sabbatical, which he has mostly spent programming at Twitter.
He works 12 hour days from his home in Maryland, keeping up with graduate students in the morning and then remotely joining his Twitter software-engineering team in California to spend the rest of the day programming.
He’s earned “street cred” from the programmers for his work on projects such as a tool that recommends which users to follow based on subject searches, but he also spends about 20 percent of his time “poking around and finding interesting things.”
The professor, who specializes in natural-language processing and data visualization, shared some of those “interesting things” in his talk Wednesday.
He told the crowd that visualizations can play a major role in helping researchers draw insights from data analysis. Visual representations of Twitter use can show how the public conversation evolves in major events, he said.
He presented an example of a visualization showing Twitter traffic during the World Cup, with the volume of tweets about participating countries presented like sound waves.
He said Twitter has used visualizations of usage data to identify countries that could be promising new markets.
In an interview after the talk, he said more companies are turning to data visualizations as a way to make sense of pages and pages worth of data, a sort of “executive summary” for the digital age.
For Mr. Lin, who will be taking a leave from Maryland for the next year to continue working at Twitter, the opportunity to step away from academe has allowed him to identify rich topics for future study and given him access to terabytes of data to analyze.
The target for professors should be to tackle questions five to 10 years down the road, he said, while leaving more-immediate development concerns to industry, which has more resources and time to devote to them.
His time at Twitter has reaped an immediate reward for one of his students, though. The visualizations he presented Wednesday were created by Miguel Rios, a graduate student at Maryland whom he brought along as an intern when he began his sabbatical last year. That internship led to a job at Twitter for Mr. Rios, who left graduate school early to take the position he’d dreamed of getting after graduating.