Knowledge Base Semantic Verification Based on Contexts Propagation

Model-Based Validation of Intelligence

Volume

Papers from the 2001 AAAI Spring Symposium

Contents

MECORI [Ramirez and de Antonio, 1998] is a method which is able to detect inconsistencies defined by integrity constraints (ICs) in a Knowlegde base (KB). The approach of MECORI is similar to that of other methods like COVADIS [Rousset, 1988] or KB-REDUCER [Ginsberg, 1988], since all of them build an ATMS-theory [de Kleer, 1986]. However, the scope of MECORI is wider than the scope of these methods, since it allows us to verify KBs expressed in a more powerful knowledge representation formalism called CCR-2 [Martinez, 1993], and it also allows us to detect a wider range of inconsistencies by supporting a flexible IC specification language.