Research action on Integration and multiple representation


BDMUL is dealing with integration of mulitsources data with multi-resolutions. In particular, the specifications of databasis are being described and methods of matching are being created, using imprecise knowledge.

The action research in integration has been active from 2005 to 2009

Leader of the research action :
Sébastien Mustière



Many geographical databases coexist to represent the same space of the real world. These bases were carried out to meet various needs (topography, navigation, town planning…) and have various levels of analysis (scale of the country, the city, the topographic details…). For example, the following design shows, superimposed at a ortho-photography, three various representations of the same crossroads, in three extracts of IGN databases with various specifications (extracted from [Sheeren 2004]). A relatively independent management of these bases raises various problems for the producer, as for the user of the data. First of all, there can be inconsistencies between the bases. Then the efforts of data capture, maintenance and update are multiplied. Lastly, it is difficult to carry out analyses combining various data with various points of view.

A possible solution for these problems is to make explicit the relations between the various elements of the databases. The passage of data organised in a unit of independent bases to data explicitly linked is called the integration of the databases. It can result in a multi-representations database, where various representations of the same phenomenon coexist and are connected between them.



The research project tries to answer to the following questions:

  • How to create a multi-representations database starting from existing databases?
    The tackled problems are then those of the modelling and the detection of correspondences between schemas, specifications and as well as the management of the connections, redundancies and inconsistencies. This integration is the main short-term objective approached by the research project.
  • How to manage and handle a multi-representations database ?
    The tackled problems are those of the propagation updates between various representations, as well as those of the visualisation and the analysis of data with multiple representation. Let us note that the aspects related to software architecture for the update are not approached by the research project.


The research project is in the continuation of work completed at laboratory COGIT on management geographical databases since ten years:

  • PhD of Thomas Devogele [Devogele 1997 ; Devogele et al. 1998] : global methodology for the integration of geographical data, and design of a tool for networks matching on various scales was developped.
  • European project MurMur : modelling of the data with multiple representation, and creation of dedicated CASE tools [MurMur 2000-2002 ; Balley et al. 2004].
  • PhD of Thierry Badard [Badard 2000] : strategy of update based on tools for automatic matching.
  • PhD of Atef Bel Hadj Ali [Bel Hadj Ali 2001] study of data quality of geomatric surfaces(quality analyses and a tool for matching surfaces).
  • PhD of David Sheeren : study of consistency between representations, approach containing automatic learning.
  • PhD of Nils Gesbert : formalisation of the databases specifications in order to guide the integration of schemas. The approach is based on the concept of ontology as a reference to the geographical concepts handled in the schemas.
  • Many studies of matching were carried out by Cécile Lemarié (from 1996 to 2000), Sébastien Mustière (from 2002 to 2005) and Eric Grosso (in 2005) [Mustière 2002].


Sébastien Mustière is working on data matching and ontologies handling for multi-representations databases.

Ana Maria Olteanu is in second year of her PhD about “Spacial data matching in taking into account of vague knowledge”. It is directed by Anne Ruas, is supervised by Sébastien Mustière, and is registered at the Marne-La-Vallée University (doctoral school ICMS).

Eric Grosso adapts GeOxygene to the particular case of the multi-represented data, in term of modelling, storage, visualisation and handling. He is working on integration web services.

Nathalie Abadie, in the continuation of the PhD of Nils Gesbert, is developping a convivial tool for handling the specifications of the geographic databases.

Christelle Pierkot is in 3rd year of PhD (EADS, IRIT, COGIT) under the direction of Abdelkader Hameurlain (IRIT-Toulouse), Anne Ruas and Sébastien Mustière (COGIT). Her subject is the metadata modelling to ease the update of the data, in a context where many actors divide, exchange, handle and update the same geographical data.

COGIT PhD on this topic


  • Supervision of a PhD CIFRE carried out with EADS/Toulouse (Christelle Pierkot) codirected by the COGIT and IRIT
  • Participation to the European project Walk On Web (IST-2-004688-STP)
IGN Web Site