Is Twitter the Ultimate Sensor?
By, Karen Braun, manager, Collaborative & Interaction Workflows & Information
Tong Sun, principal scientist and manager, Analytics and Large Scale Computing
This month last year a 5.8 magnitude earthquake struck Virginia, shaking most of the U.S. eastern seaboard just as Hurricane Irene threatened to make landfall. One minute after the earthquake struck 40,000 tweets hit the internet. Meanwhile, the National Hurricane Center got an additional 15,000 “likes” on Facebook ahead of the hurricane’s advance.
This phenomenon caught the attention of scientists around the world, including researchers here at Xerox Research Center Webster, where we’ve spent more than 50 years developing sensors for all sorts of things. In their most basic form, sensors measure something physical and convert that measurement into a signal which can be read by an observer or instrument. We asked ourselves, isn’t that what the earthquake and hurricane Tweeters were doing — collecting signals?
We’ve developed a sensor for our inkjet technology that scans billions of droplets per second to ensure they appear where they need to be on a page. This sensor helps customers predict potential problems. XRCW’s social media analytics group believes a similar approach can be used with data gathered from social networking sites such as Facebook, LinkedIn, MySpace and Twitter.
The number of people using social networking sites has nearly doubled since 2008, and the population using them has gotten older, according to the Pew Internet and Life Project.
As this “human-generated” data increases, the potential to track it (using algorithms) for real-time event detection is clear, and the applications innumerable. Researchers at Cornell University have tested the feasibility of using Twitter for real-time event detection by tracking National Football League (NFL) games, for example.
We’re thinking about how this could be applied to business processes. Think about customer care. We’re currently studying ways to detect real-time activity on social media sites that will help Xerox customers proactively manage their customer issues. For instance, by quickly detecting a network outage in NYC from incoming tweets, a timely response strategy can be deployed on several customer care channels, such as interactive social media, call-center and Websites. No need to wait for consumers to call in and report such an outage. XRCW researchers Lei Li, Tong Sun, and Wei Peng describe this system in Measuring Engagement Effectiveness in Social Media A Time-Sensitive Methodology. On another front, XRCW researchers are developing ways and tools to gain a deeper understanding of consumer behavior and interests gleaned from social conversations over time. This research could eventually help businesses better target their brand communications.
As one U.S. government official put it: “Like the weather, social media technology is constantly changing.” Xerox researchers are putting their expertise in social media, human behavior, imaging and sensors to work, to find ways to improve the processes business and government depend on.