Distinguished Seminar Speaker: Dr. Tarek Abdelzaher - University of Illinois-Urbana
This talk presents recent advances on exploiting social networks as “sensor networks”, where sources post observations on the physical world; an act we call social sensing. Social sources already volunteer to post over 500 million tweets and over 80 million Instagram photos per day, among many other social network outlets. We demonstrate that, as a sensing modality, social sensing is not unlike acoustic sensing, vibration sensing, or magnetic sensing. Appropriate signal processing algorithms inspired by physical data fusion can help exploit social media for purposes such as veracity analysis, event detection, localization, and tracking. Of particular interest are language-independent algorithms. They leverage the intuition that individual reactions to events and posts are a response to the subject matter of these events and posts. Therefore, the collective reaction pattern carries a lot of information on the nature of content posted, making it possible to analyze it without understanding the language, much the way it is possible to interpret non-verbal cues to understand conversation content. The talk describes examples of language-independent social network signal processing using Twitter and Instagram data, and presents the underlying analytical foundations and recent application results.
Monday, March 6, 2017 at 10:00am to 11:00am
Computer Science Building, 209
500 W. 15th St., Rolla, MO 65409