Citation

Rusho MA, Ahmed MA, Sadri AM. Transp. Res. Interdiscip. Persp. 2021; 11: e100420.

Copyright

(Copyright © 2021, Elsevier Publishing)

DOI

10.1016/j.trip.2021.100420

PMID

unavailable

Abstract

Active shooting, a man-made hazard, remains an unsolved challenge as communities get threatened more frequently than ever before. As a low-cost alternative to the traditional approaches of responding to such crisis, data-driven approaches can help to identify more tailored response strategies and guide towards more informed decision-making. Recently, social media platforms helped researchers and practitioners with sufficient details and coverage to understand how communities respond to natural hazards through social media interactions. However, the empirical literature does not provide any comprehensive guidance on public reactions to active shootings as observed through social media interactions. This study adopted a holistic data analytics approach to collect large-scale social media data from Twitter (~252 K tweets, 04.17.20-05.20.20). The 2020 Nova Scotia Attacks were among the major shooting events observed during this period in addition to the unprecedented experiences people were having due to the COVID-19 pandemic. This study used several natural language processing and data mining approaches (such as temporal heatmaps, word bigrams, and topic mining) to cluster the social media crisis communication patterns of active shootings and create infographics of the diverse needs, concerns, and reactions observed in the aftermath of such events. Key interactions include bailing out of shooters, shooting investigation, police response, gun violence, lessons learned from the previous school (Sandy Hook) and mass shootings (El Paso), vehicle ramming (Toronto Van Attack), mobility issues, and health concerns during COVID-19 pandemic, changes in economy and education systems. This study would allow first responders and emergency management officials to enhance the capacity of social sharing platforms and facilitate risk communication in no-notice scenarios. Additionally, the infographics could serve as a data dictionary in future active shooting scenarios to maximize peer influence.

Language: en

Keywords

Active shooting; COVID-19; Machine learning; Social media