User-generated content on social networks, such as Twitter or Facebook, offers a unique opportunity to identify public positions that make known an intention, a plan and a desire to act. The INTACT project (Detection of Engagement and Prediction of Action for Crisis Management), led by Alda Mari, research director at the CNRS Jean Nicod Institute, and Farah Benamara Zitoune, associate professor of computer science at the Paul Sabatier University in Toulouse, aims to build an automatic detection driver for these modalities for action on social networks, used during an ecological crisis or a public order crisis.
Formal and Computational Linguistics for Crisis Management
INTACT is a multidisciplinary project that brings together expertise in linguistics and computational linguistics. Its ambition is to create a pilot that analyzes the content of social networks to sort urgent and non-urgent information into relevant categories in order to raise alerts and mobilize the agents concerned by providing relevant information related to these alerts. It will be integrated into a platform hosting a machine learning tool.
The project proposes to study in depth the linguistic mechanisms allowing to predict the actions to be taken. This study will be completed by the creation of a reference corpus in French annotated according to the author's intentions to act and a system of automatic classification of messages in categories relevant to users at the Ministry of the Interior. Finally, the project aims at developing computational models that will automatically predict future actions from messages deemed relevant, based on a strong synergy between automatic language processing techniques and a deep formal semantic analysis of intention signals.
INTACT is, according to the project leaders, a pioneering pilot in France. Only the Qatar Computing Research Institute, at the University of Qatar, has a comparable tool provided by the CrisisNLP2 research team.
The CNRS "Programme de prématuration"
The "Programme de prématuration" has been set up by the CNRS since 2014 to "detect and support research projects carrying the most promising innovations".
The CNRS describes the prematurity stage as the first step in the technology transfer process. "Starting from the observation of the basic principle of an innovative idea, the concept of the technology or its application is developed and experimental demonstration is provided. The work of prematurity thus makes it possible to validate a proof of concept, to optimize a technology for a targeted application, to establish or strengthen the intellectual property strategy". This program helps finance projects, provides personalized support for researchers by dedicated professionals and prepares the strategy for commercialization.
Within the framework of this program, the INTACT project will extend its impact to industrial and health crises.
- The INTACT project
- The CNRS "Programme de prématuration
- The "Language Thought and Behaviour" team directed by Alda Mari at the Institut Jean Nicod