Management of Multiple and Imperfect Sources in the Context of a Territorial Community Environmental System

The work presented in this paper is a part of Observox, a community environmental information system for the monitoring of agricultural practices and their pressure on water resources in the Vesle basin, Champagne-Ardenne, France. The construction of Observox is the result of several research projects, and it is based on a methodology involving the actors concerned by the issue of water quality. Furthermore such a system requires the use of information provided by multiple sources which are usually imperfect. To provide the most honest indicators to the system's users, we integrate the notion of information quality by a degree of confidence. Thus we present the use of two main frameworks for imperfect knowledge management in the environmental information system, the fuzzy logic for propagating imprecision and belief functions for merging classifications.