If - Then Rules
If-then rules, often called production rules, have been by far the most widely used method of representing domain knowledge in expert systems.
A typical production rule will be of the form :
IF condition(s) THEN conclusion(s)
Two strategies for reasoning with production rules, which will be performed by the inference engine, are forward chaining and backward chaining. In both cases a inference path is sought between what is currently known and a goal.
When forward chaining is employed the known starting conditions are input into the left hand side of the production rules to generate a set of conclusions. These are checked to see if they match the desired goal, if not, these conclusions are used by subsequent rules to draw further conclusions.
When backward chaining is employed rules are selected whose right hand sides will achieve the desired goal. The left had sides are considered as intermediate goals that must be satisfied and rules to achieve these are accessed. This process attempts to find a situation where all the left hand sides are known to be true by the system.
In practise, as in MYCIN (Shortliffe 1976), both backward and forward chaining strategies can be applied.
Authored by Serengul Smith
E-mail to:
serengul1@mdx.ac.uk
School of Computing Science Middlesex University
Revised: September 1998