Multilingual retrieval of crisis-relevant information from social media

Fedor Vitiugin

Abstract: Social media is a valuable platform for sharing real-time perspectives and insights, particularly during dynamic events. Extracting relevant information from social media during emergencies can be challenging, especially when dealing with multiple languages. To address this issue, we have developed a multilingual approach that retrieves and summarizes crisis-related information from social media, providing a comprehensive solution. Our method involves experts and volunteers to address various information needs of emergency managers, allowing them to access distilled and aggregated data for validation and event monitoring. For crisis responders, targeted social media messages tailored to their operational requirements facilitate efficient work. Additionally, we propose a method for detecting and prioritizing critical help-seeking requests in multiple languages during disasters, enabling prompt and effective responses. By leveraging social media, our approach enhances emergency management by facilitating information assimilation, monitoring, and response coordination.

External IDs:dblp:phd/es/Vitiugin23