A knowledge representation based on the Belnap's four-valued logic
Yuri Kaluzhny, Alexei Yu. Muravitsky · Journal of Applied Non-Classical Logics · 1993
Introduction Much recent work in artificial intelligence has required formal techniques for working with incomplete knowledge. The new approach clearly necessitated that a distinction be made between logic and the action (i.e. two components of intelligence: the epistemological and the heuristic; cf. [MCC 69]). However, certain new inference procedures associated with the new approach (for example the familiar example of default logic) have, like classical logic, been based on an underlying consistent ontology. (Cf. [GIN 87a]: "It is precisely this `absence of information to the contrary' that makes the inference non-monotonic...") 1 Philosophers were the first to call attention to the new situation. (Cf. [RES 79]: "...we can live with the prospect of inconsistency --- not only in epistemology, but even in ontology"; and further: "The invocation of ontology here is significant... Thus if t(P ) were to be construed not as `P