DESCRIPTION OF DATA STRUCTURES AND INTEROPERABILITY
COGIT Laboratory, IGN, France
In this paper, a “data structure” means a set of schemas and rules characterizing a dataset at semantic, conceptual, logical and physical levels. Users of spatial data and processing services face three problems regarding data structures. Firstly, they must understand the structure of available datasets, namely the source structure. Secondly, they must understand the structure required by available processing services, namely the target structure. Thirdly, they must transform the source structure into the target structure. The first two problems call for adequate descriptions of structures, while the third one calls for adequate restructuration services. This paper reviews the contributions and insufficiencies of existing standards to describe structures and to assess required structure transformations.
To understand a source structure requires understanding the representation process followed by the data producer, i.e. choices for categorizing, selecting, modelling, and implementing real world entities into database objects. Some of these choices can be documented following several ISO standards: Metadata (ISO19115), Application schema (ISO19109), Spatial schema (ISO19107), Specifications (ISO19131), etc. However, there is a lack for a formal description of categorization and modelling choices, e.g. to specify what is called “a road”, and what part of a road must be represented [Gesbert, 2004]. Furthermore, descriptions of different levels of a structure should be linked together.
To understand a target structure firstly implies to know the application schema of the data processing service, including the “role” of each schema element in the process . For instance, a route processing service requires graph-structured data, and the “Arc_2” class refers to drivable sections dedicated to the computation of secondary routes [Neun et al, 2006]. It secondly requires understanding the grammar rules of the used platform, e.g. whether it supports complex geometric primitives or not. Current models for describing services like WSDL and OWL-S, currently enriched by on-going works about service chaining [Lemmens et al, 2006], can describe some structure constraints. However, the platform grammar rules and the “role” of application schema elements need further formalisation.
To transform a source data structure into a target structure, i.e. to make data and services interoperable, implies to match both descriptions. Therefore, this paper argues for an integrated, ISO-based, description model for source and target structures. As shown on figure 1, this model would give rise to a range of interesting services dedicated to data producers, service providers, and end-users, to edit, compare and transform data structures.
Figure 1 : Example of a restructuration service to feed data processing services with data. The restructuration service requires an interactive user parameterization.
Neun, M., Burghardt, D. and Weibel, R. (2006). Spatial Structures as Generalization Support Services. In Proceedings of Workshop on Multiple Representation and Interoperability of Spatial Data, pp. 6-15. Hannover, Germany.
Lemmens, R., Grannel, C., Wytzisk, A., de By, R., Gould, M ., van Oosterom, P. (2006). Semantic and syntactic service descriptions at work in geo-service chaining. In Procedings of the 9th AGILE Conference on Geographic Information Science, pp.51-61. Visegrad, Hungary.
Gesbert, N. (2004). Formalisation of Geographical Database Specifications. In Proceedings of the ADBIS Conference on Advances in Databases and Information Systems, pp. 202-211. Budapest, Hungary.