Document :

UIT-D SG02 RAPPORTEUR GROUP MEETING  Contribution  34

Titre :

Artificial intelligence-based multi-label classification of emergency communications in African languages: challenges and opportunities in the Republic of the Congo and the Democratic Republic of the Congo

Reçu le :

2026-08-13

Source :

Congo (Rep. of the)

AI/Question :

Q3/1, Q5/2

Réunion :

2026-10-07

Disponibilité :

Réservé aux utilisateurs TIES [UIT-D]

Résumé :

This contribution considers the use of artificial intelligence (AI)-based multi-label classification to improve the processing of emergency calls in the Republic of the Congo and the Democratic Republic of the Congo. Unlike a simple classification system, this approach simultaneously identifies the type of event, the level of emergency, the location, the victims, the resources needed and the relevant service. The main difficulty lies in its adaptation to the language environment, as real-world communications involve a mix of French, Lingala, Kituba, Kikongo, Swahili and Tshiluba.
The LinguaTriage (2026), TRIM-AI and Kimanuka studies demonstrate the technical feasibility of several of the components required for the automated processing of multilingual communications, without addressing the establishment of a national emergency centre. This contribution proposes a framework based on human oversight, language equality and data protection. It also presents avenues for the responsible adoption of AI and the development of digital skills, which fall under Question 5/2.