Re. Expert systems

[Geoff Hart <geoff-h -at- MTL -dot- FERIC -dot- CA>]

     Debbie Molinaro asked about sources of information on expert systems,
     but incorrectly defined the term. An expert system, simplistically, is
     a series of rules that is equivalent to a flow chart or decision tree;
     at each step, the software asks a question about the input data, and
     takes a branch to a new question based on the answer; after several
     branches, you reach a decision. The goal is to emulate the path a
     human _expert_ would take to reach a decision by formally specifying
     the _system_ of rules that the human would use.~(There are more
     modern, more complex approaches to expert systems, but the traditional
     approach has been through rule-based inference.)

     For example, to identify a tree in my yard, I might follow this
     approach:
     - If it has narrow needle leaves, it's a conifer; if broad, flat
     leaves, it's a deciduous tree.
     - (Let's assume needles:) If the needles are individually attached to
     the branches, it's a spruce; if they're grouped, it's a pine.
     - (Assume a pine): If the needles are in clusters of five, it's white
     pine; if clusters of two, red pine.
     - (Assume two): The tree is a red pine.
     (The approach is greatly simplified, but you can see the principle.)

     References? Check in your library under the keyword "artificial
     intelligence" if "expert systems" doesn't provide any matches. Byte
     magazine also reviewed the state of the art a year or two back; if you
     can't find it through their index, send e-mail to editors -at- bix -dot- com and
     they'll provide details. Good luck!

     --Geoff Hart     #8^{)}
     geoff-h -at- mtl -dot- feric -dot- ca

     Disclaimer: These comments are my own and don't represent the opinions
     of the Forest Engineering Research Institute of Canada.