Browsing by Author "LAKEHAL Meftah"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Traitement des requêtes basé sur les ontologies et l’apprentissage automatique
    (University of M'sila, 2025) LAKEHAL Meftah
    In this dissertation, we introduce a hybrid framework that integrates supervised machine learning with ontology-based semantic reasoning to optimize the execution of SQL queries in relational database systems. The relational schema is first transformed into an OWL ontology, where domain semantics and business constraints are formally represented as TBox axioms. Using historical query logs, we construct a labeled dataset that is employed to train a Random Forest classifier capable of distinguishing between simple and complex queries based on structural and execution-related features, including the number of joins, the presence of selection predicates (WHERE clauses), and observed execution times. Queries classified as complex are subsequently processed through a semantic rewriting stage, guided by inference rules encoded in the ontology. This rewriting step reduces redundant or irrelevant data retrieval, thereby improving query execution efficiency. Experimental results demonstrate that the proposed approach achieves high classification accuracy and yields significant performance gains. Collectively, these findings highlight the advantages of combining knowledge representation with machine learning techniques to enable intelligent, context-aware query optimization strategies in relational database environments.