Reports and Papers

Fedra, K. (1995)
Decision support for natural resources management:
Models, GIS and expert systems.
AI Applications, 9/3 (1995) pp 3-19.

Models and Expert Systems

Another example of possible integration is provided by quantitative numerical models and rule-based, qualitative expert systems. Expert systems can be used just as any other model to assign a value to an output variable given a set of input variables; they do this, however, by using rules and logical inference rather than numerical algorithms.

In the context of models, expert systems are often used to help configure models (implementing an experienced modelers know-how to support the less experienced user) and estimate parameters. A number of these ``intelligent front end systems'' or model advisors have been developed in the environmental domain (Fedra 1993b, 1992).

A rule-based approach can also be a substitute for a numerical model, in particular, if the processes described are not only in the physical and chemical, but in the biological and socio-economic domain. An example could be environmental impact assessment based on a checklist of problems, which can be understood as a diagnostic or classification task. A qualitative label is assigned to potential problems, based on the available data on environment and planned action, and a set of generic rules assessing and grading the likely consequences. An example of such a rule-based system for impact assessment is described below.

And finally, a model can be integrated into the inference chain of an expert system, substituting for a set of rules: from a number of antecedents (the input parameters), a conclusion (the model output) is derived, however, not by logical inference alone but by a numerical method or algorithm as one of the branches in the inference tree (Fedra 1992). Recent examples of environmental expert systems are given in Hushon (1990) and Wright et al. (1993).

The possibility to integrate models in place of rules in an expert system and at the same time use embedded rule-based components in models provides a very rich repertoire of building blocks for interactive software systems, that link policy level information with the underlying data and methods through hierarchies of methods and intermediate data (Figure 2).

The flexibility to use, alternatively or conjunctively, both symbolic and numerical methods in one and the same application allows the system to be responsive to the information at hand, and the user's requirements and constraints. This combination and possible substitution of methods of analysis, and the integration of data bases, geographical information systems, and hypertext, allows to efficiently exploit whatever information, data and expertise is available in a given problem situation.

The approach is based on a model of human problem solving that recursively refines and redefines a problem as more information becomes available or certain alternatives are excluded at a screening level. Learning, ie., adaptive response to the problem situation and the information available, and the ability to modify function and behavior as more information becomes available, is a characteristic of intelligent systems.

© Copyright 1995-2018 by:   ESS   Environmental Software and Services GmbH AUSTRIA | print page