Microblog Topic-Words Detection Model for Earthquake Emergency Responses Based on Information Classification...

Area of Science:

• Disaster Management
• Computational Social Science
• Information Science

Background:

• Social media generates vast, dynamic data crucial for disaster response.
• Efficiently extracting earthquake emergency information from microblogs is challenging.

Purpose of the Study:

• To develop and validate a topic-words detection model for earthquake emergency microblog messages.
• To construct an earthquake emergency information classification hierarchy.
• To improve the extraction of pertinent information during seismic events.

Main Methods:

• Case analysis of post-earthquake microblog data.
• Development of an earthquake emergency information classification hierarchy.
• Creation and refinement of a topic-words detection model.
• Validation using 2201 messages from the 2014 Ludian earthquake.

Main Results:

• The model demonstrated short information acquisition time.
• Overall set validity reached 96.96%; single-word validity averaged 78% (max 100%).
• New topic-words identified in different earthquake cases showed high validity.

Conclusions:

• The proposed model rapidly acquires effective and relevant earthquake emergency information.
• The classification hierarchy meets diverse earthquake emergency information needs.
• This approach enhances real-time disaster situational awareness through social media data analysis.