Abstract
Reliance on social media has become quite prevalent to initiate and assist rescue operations for humanitarian organizations. Their information needs vary greatly, from a general summary to specific needs (e.g., infrastructure damage, missing people, and injured or dead individuals). These humanitarian classes also contain several small-scale subevents that cover information from diverse dimensions. Recent methods tried to identify subevents as an event-action pair from tweets. Further, such subevents are jointly optimized with informative words, humanitarian classes using an integer linear programming framework to generate the summaries. This chapter covers the details about such subevent detection and summarization strategies. Extensive evaluation over three diverse disaster events shows that such strategy performs 6–30% better than the state-of-the-art approaches. The entire subevent and summarization framework is optimized to generate summaries in near real time.
Original language | English |
---|---|
Title of host publication | International Handbook of Disaster Research |
Publisher | Springer Nature |
Pages | 607-628 |
Number of pages | 22 |
ISBN (Electronic) | 9789811983887 |
ISBN (Print) | 9789811983870 |
DOIs | |
Publication status | Published - 1 Jan 2023 |
Keywords
- Humanitarian categories
- Subevent identification
- Summarization