| description abstract | Man-made disasters are often unexpected events that necessitate a comprehensive investigation to ascertain their cause. These investigations are critical for government agencies, insurance corporations, and other stakeholders. Social media platforms are a rich source of information, but the credibility and vast amount of data available make it challenging to extract useful evidence. Therefore, understanding the credibility of information from different groups of users on social media and its role in retrieving and filtering relevant sources is crucial. To illustrate the potential in this regard, this study examines the Notre Dame fire as a case study, analyzing tweets posted between April 15 and 25, 2019. Using a collection of lexicons, the study establishes a text-parsing and lexicon-based rule model to classify users’ opinions regarding the causes of the fire incident. Then the study characterizes the distribution of opinions between verified and nonverified users and investigates the temporal dynamics of reactions from subsets of users commenting on the event. The findings suggest that opinions from verified users were consistent with official reports, which highlights the potential value of the shared knowledge of verified users in the early stages of disaster investigation. The study further suggests that disaster responders should consider opinions from nonverified users, as they may aid investigations by identifying potential causes and providing new directions. In conclusion, this paper explores the potential of utilizing social media data as a means of supporting engineering investigations and emphasizes the importance of developing robust methodologies from crowdsourcing opinions for engineering investigations in the context of man-made disasters. | |
