ABSTRACT
The inferred knowledge obtained from a knowledge base alone could be used to reveal what is and what is not in a knowledge base. For example, if a user asks for information relating to a patient’s x-ray procedure, any response could indicate whether the patient had an x-ray or not. In general, a positive answer to a query discloses what is in a knowledge base, while a negative answer could have more than one interpretation. For example, a user could interpret a negative answer to mean that the answer is not in the knowledge base, or the user could interpret that it is in the knowledge base, but the knowledge base chooses not to reveal the correct answer to the query. These two interpretations could also depend on whether the knowledge base uses a CWA or OWA (Reiter 1977). Normally, an OWA indicates that data are incomplete or they could be somewhere else in the system and are
not restricted to a particular file or location. In a CWA, data are complete and a negative answer to a query usually indicates that the data are not present in the knowledge base. We assume an OWA; in particular, a user should not be able to distinguish, with accuracy, between the presence of facts hidden by the inference controller and the absence of facts that are available elsewhere.
