Extracting South African safety and security incident patterns from social media

Abstract

The use of social media data to gain insights into public phenomena is a potentially powerful tool. We present the use of social media data analysis, connected with crime and public safety incidents, to better understand reoccurring topics and potentially feed into an automated incident detection application. We collected a size-able dataset of Twitter posts (more than 60,000) over a 3 month period by monitoring crime and public safety related keywords linked to accounts. By splitting the data into two categories we are able to extract topics as well as compare and contrast how monitoring official crime and public safety accounts differs from monitoring individuals and organisations that may not be part of that group. Finally we discuss a prototype application, which uses social media data as well as locations to calculate metrics using potential crime and public safety related incidents.