Description
Date depot: 8 avril 2019
Titre: Gestion de la qualité de données temporelles dans un environemment streaming par une approche logique formelle.
Directrice de thèse:
Salima BENBERNOU (LIPADE)
Directeur de thèse:
Mourad OUZIRI (LIPADE)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Données et connaissances
Resumé:
Real data suffer from inconsistencies, in this purpose there have been several techniques proposed for improving the quality of data, mostly by discovering data dependencies called also Integrity Constraints (ICs) which describe the intended semantics and business rules; their violations can point out possible data errors. Prior work has focused on functional dependencies (FDs) and extensions thereof, such as conditional FDs. However, FDs capture only relationship among object in a static way and ignore the time dimension. Order Dependencies (ODs) have been introduced which subsume FDs, as they can additionally express business rules involving ordered attributes such as between timestamps and numbers. Moreover there are other errors that can be identified only through temporal functional dependencies (TFDs) which are FDs that restrict the rule on time duration. Each contribution on functional dependencies, discussed below, offers an isolated solution to one small part of the practical problem of data inconsistency and remains far from modeling the dynamic evolution of data over time and to respond within time constraints to changes in the real world. In this thesis, we address the problem of dealing with temporal inconsistency and temporal data repairing in fusion of big data using a logical approach that can permit more expressiveness in the definition of the temporal Integrity constraints between objects and enable a reasoner to take them as an input for detecting even the implicit hidden temporal inconsistencies.
Doctorant.e: Tahrat Sabiha