Social Media Analytics and Reporting Toolkit (SMART)

When a disaster or public health emergency strikes, the internet is flooded with tweets, Facebook messages, blogs, and other social media posts. If used correctly, this information can shape the way public safety agencies handle the response to and recovery from major events. However, the sheer volume of data makes it difficult for analysts to sift through and verify information in real time. Analysts need methods to monitor targeted topics on social media platforms, identify trends and anomalies, and extract and use this information to improve decision-making during disasters or crises.

To that end, the Visual Analytics for Command, Control, and Interoperability Environments (VACCINE) Center at Purdue University, a United States (US) Department of Homeland Security (DHS) Science and Technology (S&T) Center of Excellence, created the Social Media Analytics and Reporting Toolkit (SMART), a social media analysis system that provides analysts with scalable analysis and visualisation of social media posts. It allows end users to map, interactively explore, and navigate large volumes of data, topics, and anomalies that occur in real-time via social media networks such as Flickr, YouTube, Instagram, and Twitter. The system uses topic extraction, combinations of key word filters, word cluster examination, and unusual event detection to provide situational awareness and improve decision-making for time-critical tasks.

As part of the approach, end users build and customise message/keyword filters interactively and visually. The created filter methods can be arranged and adapted continually for monitoring and analysing data. SMART can enhance analysis of social media posts during a disaster in progress by comparing them to the historical information and trends of previous events, such as hurricanes and other natural disasters, terrorist attacks, and traffic. Features of the system include real-time monitoring of social media channels, extraction of trending and abnormal topics, density-based spatial clustering, message classification based on intention verbs, and task-tailored interactive message categorisation. These components are integrated into a highly interactive visual analysis workbench that allows end users to observe, supervise, and configure the methods in each analysis process. The system also incorporates visual analytic techniques to extract and visualise crowd movement patterns and trajectories using social media data at various scales of analysis to allow users to detect anomalies and outlier patterns.

SMART is a web-based tool that has been used by a number of organisations and at many national events, storms, and incidents. For instance, at the 2017 US Presidential Inauguration, SMART was used to maintain awareness of groups that advocated illegal activity, and when disruptions were coordinated online or incidents were detected, this information was used to vector public safety personnel to a specific area. Other users include: Purdue Police; Indianapolis Metropolitan Police; St. Clair County Michigan Emergency Response; American Red Cross; US C.I.S.; US Customs and Border Patrol; Intelligence Fusion Centers in Oklahoma, Tennessee, and Madison, Wisconsin; and various other law enforcement agencies.

To inquire about accessing SMART, fill out the contact form here or email universityprograms@hq.dhs.gov and/or David Ebert ebertd@ecn.purdue.edu.