Camille Bernard - Immersing evolving geographic divisions in the semantic Web. Towards spatiotemporal knowledge graphs to reflect territorial dynamics over time

Organized by: 
Camille Bernard
Camille Bernard


The members of the jury are:

  • Reviewers 
    • Nathalie Aussenac-Gilles, directrice de recherche CNRS, IRIT, Université de Toulouse
    • Christophe Claramunt, professeur des universités , Institut de Recherche de l’Ecole navale
  • Examiners
    • Thérèse Libourel, professeure émérite, Université de Montpellier
    • Sihem Amer-Yahia, directrice de recherche CNRS, LIG, Université Grenoble Alpes
    • Christophe Cruz,  maître de conférences HDR, Université Bourgogne Franche-Comté
  • Supervisors
    • Marlène Villanova-Oliver, maître de conférences HDR, Université Grenoble Alpes
    • Jérôme Gensel, professeur des universités, Université Grenoble Alpes
    • Hy Dao, professeur titulaire, Université de Genève

All around the world, the geographic divisions that cover the territories often change their names, belonging or boundaries... These changes are a clear obstacle to the comparability of data (socio-economic, health or environmental data) measured on these territories over long periods of time.  The goal of this thesis is to provide a conceptual and operational solution to this problem.

In this thesis, we present the Theseus Framework. Theseus adopts Semantic Web technologies and Linked Open Data (LOD) representation for the description of the geographic divisions, and of their evolution over time. This guaranties, among others, the syntactic and semantic interoperability between systems exchanging statistical and geometric datasets about geographic divisions, called Territorial Statistical Nomenclatures (TSN) created by statistical agencies. Theseus is composed of a set of modules which handle the whole TSN data life cycle on the LOD Web: from the modeling of geographic areas and of their changes, to the automatic detection of the changes then exploitation of these descriptions on the LOD Web, using SPARQL requests. Theseus embeds two ontologies, TSN Ontology and TSN-Change Ontology, we have designed for an unambiguous description, in time and space, of the geographic structures and of their changes over time.

The knowledge graphs generated by Theseus enhance the understanding of territorial dynamics, providing policy-makers, technicians, researchers and general public with fine-grained semantic descriptions of territorial changes to conduct various accurate and traceable analyses. The applicability and genericity of our approach has been validated by testing Theseus on three very different official geographic divisions.
Finally, we also present a set of predefined queries to query the created knowledge graphs immersed on the LOD Web, in order to highlight the types of territorial changes characterized by the TSN-Change Ontology.