A GRAPH-THEORETIC APPROACH TO EXPERT-SYSTEM TESTING

Logic paths play an important role during the testing of an expert sys tem, as the number of test cases to be used is determined by these pat hs, and they are used for testing the system under consideration durin g implementation. This paper proposes an algorithm based on a graph-th eoretic clustering concept to determine the logic paths in an expert s ystem. It also allows the testing of the expert system during the vali dation and verification phases for defining the optimal number of test cases. The proposed algorithm is efficient and compact, and guarantee s the consistency of the underlying system during its implementation p hase.