Representation and use of the knowledge

Abstract

The representation of the knowledge is the main task when developing an expert system. Although, for that matter, this problem is also encountered in other areas of research such as language comprehension, robotics or intelligent data bases. In fact as the representation of the knowledge is a formalism serving as medium for the phenomena being studied, the inference capacity is powerfully influenced by the choice of representation. This choice is made initially by deciding on either procedural representation or on declarative representation. A procedural representation makes clear the relationship between the elements of knowledge whilst the declarative representation attempts to express knowledge in the form of independent ‘granules’ and leaves to a reasoning mechanism the grouping of the knowledge elements for making deductions (Pinson, 1981). The details of the procedures used to represent knowledge will not be gone into. It leads to the writing of programs which are difficult to modify and extend as the knowledge develops. Schematically, if one is primarily interested in declarative representation, whether pure as in production rules and the semantic network or mixed as in frames, there are two ways of approaching the problem.