Fault detection algorithm for logic programs

摘要

A fault detection algorithm for logic programs is presented which is able to find several faults in one pass, in comparison with other techniques such as that of Shapiro1. In the context of software reliability, Angus et al.2 show a non-monotonic reliability profile where reliability curves are given by the number of faults encountered versus time. One explanation of this is the later discovery of faults which were originally masked by other faults. The system is able to derive sets of faults at different levels or branches of the proof tree, and so exploits this explanation of non-monotonic reliability in the debbugging process.