Partner und Internationale Organisationen
(Englisch)
|
Star Informatic et Université Libre de Bruxelles (B), Institut Géographique National et Cemagref-Grenoble (F), EPFL (CH)
|
Abstract
(Englisch)
|
The MurMur project aimed at defining a new approach for database description and manipulation that would fully support the need for defining and storing multiple perspectives of the same information. The MurMur project focused in particular on geographical databases, coping with the following two situations: multiple representations due to multi-scale(s) databases and multiple representations due to time-varying information.
The approach followed by the project has been to use a conceptual perspective for coping with the problem of multiple representations. From a modeling perspective, conceptual means that the effort is made to represent a real-world phenomenon in terms as close as possible to the final users of the system, where all technical details concerning the implementation platform are excluded. Therefore, the first concern was to extend an existing conceptual model with multi-representation capabilities. MADS, a conceptual spatio-temporal model developed by UNIL and EPFL, served as the starting base.
The specific multi-representation features that were addressed by the project were driven by a requirements analysis from a Cartographic case study provided by IGN and a Risk-Management case study provided by Cemagref.
The first main result of the project has been the extension of the MADS model with multi-representation capabilities. The extension was carried over as an addition of yet another modeling dimension to the model, orthogonal to the existing spatial and temporal dimensions. This choice facilitated design decisions and maximized the expressive power of the model.
Next, three complementary tools were built. A visual schema editor, allowing users to define the conceptual model of an application. A schema translator, transforming a high-level conceptual schema into a low-level conceptual schema close to the targeted implementation platform. Finally, a wrapper, translating the low-level conceptual schema into the specific syntax of the target platform. There are as many wrappers as target systems. The chosen platform was Oracle, although some tests were also done for the MapInfo and ArcInfo platforms. All these tools were developed using an iterative lifecycle, where each cycle comprised extensive tests by the users.
The definition of an associated manipulation language lead to the second stage of the MurMur project: the development of the corresponding query facilities. MurMur has developed such facilities at the conceptual level, and materialized the approach into a visual query interface comprising a query formulation tool and a multidimensional query viewer. These tools for defining a query and visualizing the result of query execution were made operational by the development of a query translator and a query wrapper for an Oracle DBMS.
Finally, two advanced services were developed: intelligent zooming and multidimensional views. Intelligent zooming consists in exploiting all the representations for the same object(s) at different scales and display the representation that corresponds better to the resolution at which the user is looking at a map. Complex issues had to be solved since a set of objects at one representation may correspond to a different set of objects in another representation. Multidimensional views, originally called time-travel, consists in exploiting the temporal data sets for showing the evolution of real-world phenomena. The name change was motivated since the same approach is used for advanced visualization features exploiting the other modeling dimensions. Both advanced services were introduced into the query language and the corresponding tools.
|
