• Funding : Artois
  • PhD defended on :
  • Dec 5, 2025

This thesis lies at the intersection of qualitative spatial and temporal reasoning and data mining. It focuses on the extraction of knowledge from complex data, aiming to develop a formal and methodological framework for discovering patterns and insights that are both relevant and interpretable, a goal that remains challenging for classical data mining approaches.

The work is structured around two main axes. The first addresses the mining of qualitative patterns : it formalizes the concept of qualitative patterns in qualitative databases, defined as collections of qualitative constraint networks, and introduces tailored methods for their extraction. The second focuses on qualitative clustering, proposing techniques grounded in the semantic coherence of qualitative constraint networks. These contributions combine both declarative and algorithmic approaches and are supported by experimental evaluations that demonstrate their feasibility and relevance.

Committee

  • Salima BENBERNOU, Professeur des Universités, reviewer
  • Maroua BOUZID, Professeur des Universités, reviewer
  • Jean-François CONDOTTA, Professeur des Universités, supervisor
  • Souhila KACI, Professeur des Universités, examiner
  • Daniel LE BERRE, Professeur des Universités, examiner
  • Yakoub SALHI, Professeur des Universités, supervisor
  • Michael SIOUTIS, Professeur Junior, examiner