S108 Potential for Predictive Analysis Using Social Media

Sunday, 22 January 2017
4E (Washington State Convention Center )
W. Logan Downing, Purdue Univ., West Lafayette, IN; and D. Niyogi

     Social media, in conjunction with the internet at large, has developed into the most data rich environment in all of history. Social media services, such as Twitter, combine location data alongside of a user’s personal comments about what is happening in their lives. In the case of Twitter these comments are referred to as “Tweets”. Twitter’s partnership with Foursquare has, as of recently, greatly improved the location data available through Twitter. It is the intention of this study to evaluate the legitimacy of Twitter data as either a predictor of severe weather events or a response.

     Methods of data collection will be discussed as well as suppositions about useful architecture on Twitter’s side that would increase the plausibility of analyzing “Tweets” for real-time events. Other social media platforms will be briefly touched upon and their potential value for severe weather information will be evaluated, though not thoroughly. An analysis of previous works of this type will be performed alongside of this study and included to continue to build the progress of the scientific community at large in its endeavors to harness social media.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner