Entailment Calculus as a Logical Tool for Reasoning Rule Generation and Verification

A common inadequacy of current knowledge-based systems is that they cannot reason about those situations and/or problems that have not been considered by their developers. A major cause of this inadequacy is that the system cannot automatically generate new and valid reasoning rules from those existing rules and facts that are programmed or inputted in the systems by their developers or users. On the other hand, the lack of formal and rigorous verification technique is a major reason of that many experimental knowledge-based systems cannot be reliably used in practices for solving business and industrial problems. Both reasoning rule generation and reasoning rule verification require a domain-independent fundamental theory to underlie justifications. This paper investigates the issues of reasoning rule generation and verification in knowledge-based systems from a logical viewpoint. The paper points out that various existent logic systems fail to serve as the fundamental theory underlying reasoning rule generation and verification because their logical theorems include a lot of implicational, conjunction-implicational, and disjunction-implicational paradoxes. As a solution for the problem, the paper suggests that we should use an entailment calculus based on a paradox-free relevant logic as a logical tool for reasoning rule generation and verification in knowledge-based systems.