Stakeholder consultation for Baysian Decision Support systems in environmental management

Abstract

Environmental management is a field where the number of variables, their interactions and feed-back loops require tools to integrate information, build scenarios and support decision-making. Among the various kinds of Decision Support Systems, Bayesian Belief networks have proven to be quite useful as they can integrate quantitative information as well as qualitative expert knowledge. Beyond knowledge integration, the consultations also help clarifying what is at stake, how variables interact and what are the conflicting interests. Gathering expert knowledge requires consultations, with a particular approach constrained by modelling requirements. In this paper based on our experience we focus on the technical aspects of such stakeholders' consultation. We describe in detail the steps of the consultation, we analyse the methodology (selection of stakeholders, collective building of a model structure, probabilities elicitation, etc). Then we review the possible pitfalls and problems encountered in the process. We ultimately propose generic guidelines for stakeholders consultation in view of building Baysian models for environmental management.