Social media Triggers for Alerting and Response (STAR)

Information present in social media is valuable to emergency response organizations which have a need to maintain situational awareness about the events and activities in a region of interest. Approaches to using social media to inform emergency personnel in rare and large-scale events such as earthquakes and hurricanes have been studied. However, the emergency response community has little experience with use of social media to improve the information available to first responders in more routine emergencies. The research proposed here applies Natural Language Processing and Machine Learning algorithms to the problem of extracting the signal of emergency incidents from social media data. In Phase I, the DAC team will identify a set of emergency incident types whose social media profile allows for incident detection, alignment with CAD incident data, and extraction of response-relevant information. At the conclusion of Phase I we will understand the tradeoffs between automated processing and manual analysis of social media and will be able to quantify the benefits social-media-sourced information can bring to the first responder. We will develop a prototype system that will demonstrate the feasibility of our approach, and define the requirements and touchpoints for integrating social-media algorithms into CAD systems. The technology developed under this effort can be transitioned into commercial emergency dispatch systems, and easily repurposed to address a variety of law enforcement and intelligence analysis scenarios. This work will be developed in partnership with the FDNY.