Extracting X (formerly Twitter) information to improve relief logistics: social media analytics to the rescue

Purpose During sudden disasters, information is needed on victims’ needs and locations, resource availability, and first responders’ status to improve relief logistics; this study aims to extract this critical information from X (formerly Twitter). Design/methodology/approach Using social media analytics and the capture-understand-present approach, the authors rank users and their posts to identify relevant data to support humanitarian operations. The pipeline integrates text preprocessing, dictionary-based matching with three domain dictionaries (logistics, medical, political), confusion-matrix evaluation, sentiment analysis and a co-occurrence word network. Two past disasters serve as case studies. By interviewing users, the authors also get insights into motivations to post. Findings The methodology ranks relevant posts and users, identifying potentially actionable data. The authors observe phenomena in digital social networks during a disaster, such as misinformation, agendas to gain followers, political opinions disguised as facts and specific needs of underrepresented users, whose data is often missing. Practical implications Social analytics can identify key stakeholders in situ and improve relief logistics; first responders should directly contact top-ranked users (97 interviewed here) to surface needs invisible at the network level. Misinformation and political views contribute to irrelevant posts. Originality/value Previous research analyzes X data at the network level, which can lead to biased decision-making in relief logistics and inequality in aid deliveries. This study contributes a node-level ranking that operates without popularity metrics and triangulates the algorithmic results with interviews of the top-ranked users.