Digging Communicative Intentions: The Case of Crises Events

2026-04-19 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Natural Language Processing · Depth: Expert, short

Summary

A study by Benamara, Mari, Meunier, Moriceau, Moudjari, and Tinarrage explores the correlation between speech acts (SA) and urgency in social media content during crisis events, aiming to assist rescue teams. The research introduces a two-layer annotation scheme for speech acts at both tweet and sub-tweet levels. It also presents a new French dataset comprising approximately 13,000 tweets, annotated for both urgency and SA, covering various crisis types like storms, building collapses, and explosions. The authors conducted a detailed analysis of these annotations, focusing on the correlation between SA and message urgency, intentions to act (e.g., human damages), and crisis types. Finally, deep learning experiments were performed to detect SA in crisis-related corpora. The findings indicate a strong correlation between SA and urgency annotations, particularly at the sub-tweet level, marking a significant step toward SA-aware NLP for crisis management.

Key takeaway

For research scientists developing NLP systems for emergency response, understanding the strong correlation between speech acts and message urgency in social media is crucial. You should consider integrating speech act detection at both tweet and sub-tweet levels into your models to improve the accuracy of identifying critical information during crises. This approach can enhance the ability of rescue teams to prioritize and respond effectively to urgent situations.

Key insights

Speech acts in social media correlate strongly with message urgency during crisis events.

Principles

• Speech acts reveal intentions and preferences.
• SA patterns differ in crisis contexts.

Method

A two-layer annotation scheme for speech acts (tweet and sub-tweet) was applied to a 13K French tweet dataset, followed by deep learning for SA detection and correlation analysis with urgency and crisis types.

In practice

• Develop SA-aware NLP for crisis management.
• Identify urgent messages via SA analysis.

Topics

• Communicative Intentions
• Speech Acts
• Crisis Management
• Social Media Analysis
• Urgency Detection

Best for: Research Scientist, AI Scientist, NLP Engineer, Data Scientist

Related on AIssential

• Implicit and Indirect: Detecting Face-threatening and Paired Actions in Asynchronous Online Conversations — Detecting face-threatening actions in asynchronous conversations requires handling ambiguity and multiple interpretations.
• Translating Under Pressure: Domain-Aware LLMs for Crisis Communication — Domain adaptation and preference optimization can create readable, adequate crisis translations from limited data.
• Interpretable Crisis Behavior Analysis Using Mobility and Social Media Data — Integrating mobility and social media data reveals interpretable, actionable crisis behavior patterns.
• From Codebooks to VLMs: Evaluating Automated Visual Discourse Analysis for Climate Change on Social Media — VLMs can effectively analyze social media images for climate discourse, even with moderate per-image accuracy.
• Multimodal Support System for Disaster Risk Management — NOAH is a multimodal AI system for disaster management, integrating NLP, speech, and vision via WhatsApp for Brazilian cities.
• The Register Gap: A Meaning Intelligence Framework for Nigerian Public Discourse — AI systems fail on Nigerian discourse due to context failure, not translation, requiring deeper meaning intelligence.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.