In addition to generating huge amounts of information for the infosphere, social media are providing clues to behavior that analysts are tapping for marketing predictions. Seemingly innocuous behavior is revealing valuable information about group and individual behavior. Bernardo Huberman, senior fellow and director, Social Computing Lab, HP Labs, told the MILCOM 2010 Wednesday keynote panel that experiments have shown how this data can predict marketing success. One test allowed experts to predict the box office revenues of movies before they opened, based on tweets about the movies. Ed Leonard, chief technology officer, DreamWorks Animation, added that he and his colleagues are able to predict precisely how a movie will fare over the next 10 years based on just a few hours of box office reports. Huberman described a concept known as sentiment analysis, in which looking at a piece of text can illuminate the opinion and positive or negative feeling of the writer. This exercise cost only a few dollars, he added. His company also developed a technology that allows a small group of 20 people to predict future events. This effort measure their risk attitudes, which has broad applications. In the retail market, their attitude may determine whether people are likely to buy a laptop or not. Russ Daniels, vice president and chief technology officer, cloud services strategy, Hewlett-Packard Company, described how just knowing locations-geolocation data is being revealed by more devices and applications-can reveal patterns of behavior. Knowing the location of actions over time allows a user to understand intent. By capturing data such as where people are hailing taxis in a city in a given time frame, marketers can offer services tailored to that activity.