Blog post

EDUBASE DRC: an AI RAG assistant to improve data transparency in the education sector

Ⓒ IIEP-UNESCO/Imageo

Autor(es)

Emmanuel Vaslin, International Technical Expert (ETI), Expertise France (AFD Group), Kinshasa, DRC.

Emmanuel Vaslin

International Technical Expert (ETI), Expertise France (AFD Group), Kinshasa, DRC

EDUBASE DRC was born in November 2024 from a simple but worrying observation: in the Democratic Republic of Congo (DRC), data relating to the education sector are often scattered, inaccessible or unreliable, which limits their contribution to improving the system. While the EDUBASE assistant will not provide a definitive solution to all problems related to the collection and use of data in the DRC, it provides a practical response to a fundamental need: to centralise and simplify access, for as many people as possible, to key information on education.

The excessive dispersion of data poses a number of challenges. They are often collected in isolation by different players - ministries, NGOs, technical and financial partners - without coordination or integration into a unified system. Their centralisation at the national level limits their use at the local level. And finally, in the absence of robust monitoring and evaluation systems, the data are often obsolete, incomplete or inconsistent.

Even when data do exist, they are not always accessible to anyone beyond experts and managers in the directorates. Despite the efforts made with the new institutional websites (the e-SIGE service, for example, of the Ministry of National Education - New Citizenship), statistical directories and other data in the DRC are still too often out of reach1

EDUBASE, an IA RAG assistant

This lack of data accessibility is behind EDUBASE, an Artificial Intelligence (AI) RAG (Retrieval-Augmented Generation) assistant, developed by the author. It combines two technologies: information retrieval and content generation.

EDUBASE begins by searching for data in the corpora imposed on the machine by its creator, in this case, RESEN III2, as well as interactions (3,602 prompts and responses sent between its launch in November 2024 and the present day) between the algorithm and its users. EDUBASE is thus strengthened by relying on a set of corpora that is enriched and refined according to the needs expressed and the documents shared, then generates adapted responses using a natural language model (LLM, in this case, the GPT 4o model from Open AI).

In a context such as that of the DRC, this RAG technology offers significant advantages: it provides access to local and diversified information by retrieving data from specific sources provided to the chatbot, which helps generate relevant and contextualised responses, as illustrated in the box below

Sidebar. Experiments

The question "What is the share of salaries in the education budget of the DRC?" was put to three generalist AIs (GPT-4o, Le Chat and Deepseek) and EDUBASE. This experiment largely demonstrated the superiority of an AI assistant built using RAG methodology. To answer this question, EDUBASE RDC relies on RESEN III, published in August 2022, and provides concrete figures for two consecutive years. It adds a factual comment highlighting the small margin left for infrastructure, teaching materials and continuing education. Edubase could become even more effective if its corpus were enriched by documents other than RESEN III, supplemented by interactions between the bot and users.

What's more, the assistant can be accessed via WhatsApp, guaranteeing 24/7 access even for users without a broadband connection. This low-tech solution requires no complex or costly infrastructure.

Impact on transparency and accountability

Edubase can strengthen three types of accountability by providing accurate, accessible and usable information for the various players in the education system:

  1. Bottom-up accountability: Edubase enables ministries of education and funding bodies to base their decisions and justifications on reliable data. For example, a ministry can show donors trends in enrolment and completion rates, the distribution of budgets, or the impact of reforms on the quality of education.
  2. Horizontal accountability: The exchange of information between schools, NGOs and parents' associations is facilitated to improve the management of the education system.  Example: An NGO committed to reducing educational inequalities can use it to compare the performance of urban and rural schools, and work with the relevant authorities to adapt their interventions.
  3. Top-down accountability: Edubase contributes to transparency by making key data on education available to citizens and the media, thereby encouraging citizen involvement and oversight. Example: Parents can consult the success rates of local schools, the number of pupils per class, the teaching resources available and the actual presence of teachers, enabling them to call for improvements.

All in all, by centralising relevant data and making it accessible, Edubase promotes more transparent, efficient and evidence-based educational management.

You can test EDUBASE here

1. The query "RESEN III RDC" does not lead to the report on the 30 results of the first 3 pages of the Google search engine consulted on 13 February 2025.
2. State Report on the National Education System.

Deje su comentario
La dirección de correo electrónico no será publicada.
5 + 7 =
Resuelva este simple problema matemático y escriba la solución; por ejemplo: Para 1+3, escriba 4.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Manténgase informado

Inscríbase al boletín de ETICO para recibir las últimas noticias

Envíe su contenido

Ayúdenos a aumentar nuestra biblioteca compartiendo su contenido sobre la corrupción en la educación