Explanation in ecological systems

The use of expert systems in ecology can benefit greatly from recent advances in explanation generating techniques. These techniques can facilitate the construction of expert systems and increase manager confidence. The objectives of our research work are to design and construct expert systems software that can produce enhanced explanations in expert systems for use in the ecological domain. Our first effort was the identification of the functional requirements of explanation that were relevant to ecologists. Three needs were identified: demonstration of the relationships between the specific kinds of uncertainty modeled in the ecological reasoning process, quantification of the effect which these kinds of uncertainty have on the conclusions of the system, and identification of “critical states” in this reasoning process. The second effort was the design of efficient software modules for integrating these requirements into existing expert systems. We present an explanatory technique which can analyze and present the model of the reasoning process, used by expert ecologists in a way which can be easily related to practicalities such as the allocation of priorities, and, a mechanism to identify and analyze points in this reasoning process that become relatively critical during a consultation.