Social Media Use in Disaster Response: Empowering Communities Resilience

Description

Social media metadata is employed to investigate the use of two social media platforms, Facebook and X, during the 2021 flood in Belgium, both in the immediate aftermath of the crisis and in the short-term. First, Facebook data is employed to analyze the activities of Facebook community groups established post-crisis over a six-month period. Then, the tweets related to the 2021 flood in Belgium are analyzed. The analytical framework employs (1) a social media data-driven quantitative and qualitative text analysis using topic modeling and sentiment analysis. The findings highlight the different roles played by the two social media platforms. Facebook served as an effective platform to mobilize and organize local communities for immediate and practical support, while Twitter served as a platform for broader global engagement and advocacy. The convergence of results from diverse data sources provides comprehensive insights into the effectiveness and challenges of leveraging social media for community resilience in the aftermath of disaster events.

Steps to reproduce

Employing web data crawling software, posts from Facebook and Twitter were gathered. First, all posts from five of the addressed Facebook groups and pages were extracted. For the sixth group with broad coverage, keyword searches were used to scrape posts. Posts, along with associated comments, dates, and metrics such as reactions and shares, were collected from the period spanning from July 15th, the date of the crisis, to December 31st, when activity ceased for most of the Facebook groups involved. The number of scraped posts per page ranged from 43 to 355. The number of members of public groups ranged from 26,500 to 266, and the total number of Facebook page members varied from 2,100 to 163. It is worth mentioning that the aim was not to compare the activities of these groups or the most and least successful ones in mobilizing people, but to precisely address the digitally mediated practices in a broad sense and their contribution to building the community and enhancing resilience. Therefore, all 865 posts and 1,164 comments from the different groups and pages were combined into one textual database. The following step involved running an advanced search on Twitter to scrape tweets with two hashtags combined as follows: “#Belgium” or “#Wallonia” and “#flood” or “#floods” or “#flooding,” or “#flood2021.” Posts in English, French, or Dutch were sought. After excluding duplicates and irrelevant posts, the total number of analyzed tweets for the time period between mid-July and December 2021 was 1,130, of which 912 were posted between July 15 and 31. All metadata associated with the tweets, including each user’s country and profile, were retrieved. Facebook posts were in French and Dutch, whereas tweets were also available in English. The textual data were translated into English using the Google Translate function in Google Sheets. The data were then converted into a data frame and subsequently into a corpus using RStudio software. The next step involved preprocessing the text corpus using the tm package to eliminate punctuation, numbers, stop words, and specific words that did not carry any relevant information for the analysis, while also converting the text to lowercase and stripping white space.

Institutions

Categories

Funding

Licence