Géraldine del Mondo, PhD in Geomatics
Her research work (led at the French Naval Academy Research Institute - IRENav) is entitled "A Spatio-Temporal Graph-Based Model for the Evolution of Geographical Entities"
The jury, composed of:
-* Pr Jacques Tisseau, Professor at the LISyC, Laboratoire d'Informatique des Systèmes Complexes and Head of the ENIB - Ecole Nationale d'Ingénieurs de Brest,
-* Pr Jérôme Gensel, head of the STEAMER team at the LIG - Laboratoire d'Informatique de Grenoble,
-* Dr Florence Le Ber, researcher at the LHyGeS - Laboratoire d'Hydrologie et de Géochimie de Strasbourg,
-* Dr Roland Billen, Lecturer at the Geomatics Unit of the University of Liège, Belgium,
-* Pr Christophe Claramunt, Head of the IRENav and PhD supervisor,
-* Dr Rémy Thibaud, Head of the GIS Research Group at IRENav and co-supervisor,
-* Dr John Stell, Senior lecturer at the Institute for Artificial Intelligence and Biological Systems of the University of Leeds, United Kingdom,
-* Pr Gérard Ligozat, Professor at the University of Paris Sud, France,
awarded her the title of PhD of the Western Brittany University in Geomatics, with a summa cum laude distinction.
Space and time are fundamental concepts needed for the analysis of the evolution of spatio-temporal entities. In order to model these evolutions, the changes and relations that modified and characterise entities must be identified. These changes are especially qualified by modifications of their position, spatial footprint, or their identity.
Spatio-temporal modelling requires the development of solutions of representation which allow for the identification and study of processes involved, of theirs causes and theirs consequences. The fact that a formal representation and computing of a phenomenon leads to a simplified representation of the reality, implies that this modelling should keep the level of details available and/or aimed at for this description. Ideally, it should be possible to study a phenomenon at several levels of granularity in space and time. If numerous approaches have been proposed for the modelling of the evolution of Entities in spatio-temporal phenomena, none has a structure which fully takes into account the semantics of these applications.
This research proposes a spatio-temporal graph model which allows for the characterization of main properties of spatio-temporal evolution. Several concepts are distinguished, such as the notions of identity and relations, through the spatial and time dimensions. For a given time, we categorize Spatial and filiation relations, and through time, temporal filiation and spatio-temporal relations. The emerging graph structures characterize the evolution of a set of spatial entities and permit to discover new properties.
Graph manipulation functions are developed and applied to the spatio-temporal graph model. These functions identify generic processes (e.g. life and death of an entity) or processes linked to a specific application and its semantics. In order to combine several sources of information in a single graph, graph join functions permit the integration of several graphs in a unified representation. The properties of these graphs and the role of their entities are studied by a qualification of the different types of routes that link these entities.
The consistency of the spatio-temporal graph model is verified by a constraints-checking algorithm. We make a difference between domain constraints belonging to the graph model, and semantic constraints which depend on a specific application. An extension of the modelling approach is achieved from a structure based on bigraphs, that facilitates the explicit representation of a phenomenon according to different levels of spatial granularity.
Keywords: spatio-temporal modelling, graph theory, constraints specification