Back to activities
GERAD seminar

Large-scale optimization methods for logical reasoning: A novel perspective


May 15, 2024   11:00 AM — 12:00 PM

Maryam Daryalal HEC Montréal, Canada

Maryam Daryalal

Presentation on YouTube.

Description Logics (DL) are formal languages for knowledge representation, enabling structured ontology creation and logical reasoning. This talk presents an optimization-based framework for enhancing reasoning in DL by employing mixed-integer programming methods. The framework exploits a novel approach of mapping DL axioms to a set of inequalities, enabling the use of advanced optimization techniques. The integration of column generation and branch-and-price algorithms addresses the complexity of the resulting inequality system, offering a scalable solution for handling large ontological datasets. In a preliminary set of experiments, the framework's performance is evidenced by its application to the ontology of the Canadian Parliament, where it outperforms traditional reasoning methods. This work could mark a significant advancement for the Semantic Web community by merging ontology reasoning with sophisticated optimization methods to meet the challenges of semantic data processing.

Olivier Bahn organizer


Hybrid activity at GERAD
Zoom et salle 4488
Pavillon André-Aisenstadt
Campus de l'Université de Montréal
2920, chemin de la Tour

Montréal Québec H3T 1J4