twitter and disasters

social resilence fingerprints

One of my current research interests revolves around using social media data to better inform decision making in natural disasters, with a specific focus on hurricanes. Onf of my 10-year research goals is to develop methods to predict how a community will respond to a major disaster prior to the event and with no historical information about the imapct. For example I want to know how Portland Maine will respond to a Katrina-like hurricane even though Maine has never experienced a major storm.

To do this, current work is using social media data. We have found (publication under review) that the response of different types of disasters are unique in the subsequent social-media response of the affected. Current research is focused on utilizing this data in real-time to predict the impact of major events in advance.

Slides from a recent talk given to Purdue's IE590: Predictive Modeling class all about how to parse, clean, and construct mathematical objects from Twitter data.

publications

* IEEE Access, describing the SRF methodology and validating it