Quality-aware, Metadata-based interoperability for Environmental Health
Résumé
This paper addresses impedance mismatch problems, which occur when two information systems are plugged for working together. Such problems are particularly crucial for environmental health, where business activities require data sets from several heterogeneous and distributed systems, within autonomous organizations. We first introduce an impedance mismatch metaphor, between Information Systems, in particular between geographic data sets, developing analogy on scale problems, semantic misfits, space incompleteness, vector-to-raster issue. Then we propose an approach for a quality-aware interoperability, which is based on a step-by-step use of metadata. It focuses on early requirements, in terms of three steps: (i) existence, of relevant data, (ii) quality, sufficient for fitting the intended uses, (iii) contents, for a consistent and effective use of the data sets. We present an integrated view of these three steps, which is driven by the application requirements. A global architecture is associated with this view, by coupling a reasoning system and an integration system (mediator). The top level of this architecture is detailed, in terms of: (i) application ontology, (ii) specification languages such as Description Logics and OWL-DL, (iii) implementation choices, with Protégé and Racer systems. Finally, a simplified example illustrates the approach, and outlines several questions whose understanding can be helped by the analogy with impedance mismatch problems.