Experts from PREDICT-Services (risk management specialists and partners in the RéSoCIO project) monitoring social network activity
© BRGM - Samuel Auclair
Some natural disasters affect large areas in just a few hours (flash floods or torrential mudslides, as at La Bérarde in Isère on 20 June 2024) or just a few seconds (earthquakes). They plunge those involved in crisis management (prefectures, town halls, emergency services, etc.) into uncertainty as to the exact state of the situation. It takes a long time between the moment information is fed back from the field and the moment it is analysed and used to make decisions.
Human sensors in the field thanks to social networks
Social networks can be invaluable allies in remedying this situation, and in helping those involved to understand the situation on the ground more quickly. By acquiring eye-witness accounts from victims and bystanders via these networks, useful real-time information can be gleaned.
This means developing digital tools capable of automatically analysing social networks, filtering them and extracting potentially useful information.
Geolocation of tweets reporting in real time the earthquake that devastated the commune of Teil in Ardèche on 11 November 2019
© Suricate-Nat
AI powered by X/Twitter data
This was the conclusion reached in 2020 by scientists in an interdisciplinary team comprising BRGM, IMT Mines Albi, Université Paris-Dauphine and PREDICT-Services, which led to the launch in 2021 of the project ReSoCIO supported by the French National Research Agency.
The RéSoCIO project used Twitter's (now X) API (application programming interface) to automatically collect data, which was then analysed using artificial intelligence (AI) techniques.
Less and less accessible data
Until the beginning of 2023, Twitter allowed anyone to use some of its data completely free of charge, particularly for academic research.
Since its takeover by Elon Musk, Twitter, now X, has undergone a drastic transformation of its business model. This resulted in the outright withdrawal of its free API, replaced by paid solutions at prohibitive prices.
At a time when social platforms, like Meta, Facebook's parent company, are boasting of "data for good" policies to encourage initiatives that are useful to society, it seems necessary to guarantee privileged access to social network data not only to researchers, but also to developers of tools that have a strong social impact but that are not necessarily viable financially.
