ABSTRACT
AGASSISTANT is a new generation expert system builder for personal computers in the domain of agriculture. The AGASSISTANT advisory system consists of a set of modules accessible through menu-driven screens. While many agricultural production processes are inherently complex, a subgroup of production processes can be described adequately and converted to an appropriate knowledge base for use with AGASSISTANT. Agricultural scientists often provide advice to agricultural managers on the basis of an evaluation of their incomplete knowledge and experience. Many underlying principles of agricultural sciences can be experimentally measured in field experiments. The chapter explains how individual rules are evaluated and how uncertainties are propagated through a hierarchical knowledge base. It also explains the control structure of the system, i.e., the method by which the advisory system generates questions. While the Advisor module is most apparent to the user, at the heart of the AGASSISTANT is a flexible inference engine that runs on a compiled set of rules.
