Towards an Ontology and Knowledge Graph-Based Recommendation Approach Enhanced with Large Language Models for Civil Service Recruitment in Burkina Faso

---

Résumé

Recruitment processes in the civil service often face challenges related to the management of competencies, complex regulatory rules, and the lack of explainability in selection decisions. To address these issues, we propose a hybrid recommendation approach that integrates a modular core ontology, a dynamic knowledge graph, and large language models (LLMs) for automated knowledge extraction. This paper focuses on the overall system architecture and the development of the first component, COREM-CiS (Competence Ontology for REsource Management in the Civil Service), a modular ontology aligned with the Interministerial Directory of Jobs and Occupations (RIME) and tailored to the Burkinabé civil service. COREM-CiS provides a structured representation of professions, skills, and personnel data, forming the foundation for future integration with reasoning engines and LLMs. The proposed approach lays the groundwork for an intelligent, explainable, and scalable recruitment recommendation system in the civil service domain.

Mots-clés

Knowledge engineering , Large language models , Systems architecture , Knowledge graphs , Ontologies , Resource management , Personnel , Sustainable development , Recommender systems , Recruitment