Algorithmic Debugging and Hypothetical Reasoning

This paper presents a generalization of Shapiro style algorithmic debugging for generalized Horn clause intuitionistic logic. This logic offers hypothetical reasoning and negation is defined not by failure but by inconsistency. We extend Shapiro's notion of intended interpretation, symptoms and errors and give formal results paralleling those known for definite clauses. We also show how a corresponding diagnosis module for RISC - a logic programming system for generalized Horn clause intuitionistic logic - can be defined by meta interpretation. In contrast to Shapiro's PROLOG modules ours work independently of the specific computation rule that in RISC may be specified by the user.