Application of Clustering Analysis to Coverage Testing for Large KBS
Valerie Barr, Mala Mehrotra · 1998
TRUBAC (Testing with RUle-BAse Coverage) is rule-base testing method which evaluates the effective-ness of test sets in covering relationships within the rule-base and identifies sections of the rule-base which have not been tested. This facilitates selection of ad-ditional test cases and can lead to a more accurate prediction of system reliability in real use (Barr 1995; 1996; 1997; 1998). However, it is not clear that the ap-proach used in TRUBAC, which involves constructing a graph representation of the rule-base, will scale-up to larger systems. MVP-CA(Multi-ViewPoint Clustering Analysis) Tool (Mehrotra & Wild 1995), on the other hand, is an analysis tool to semi-automatically parti-tion a rule-base system into clusters of related rules so as to expose its semantic underpinnings. MVP-CA tool