Symbolic normalized acquisition and representation of knowledge (Q1077171)

The authors present a declarative language - SNARK - and the associated inference process. SNARK is based on first-order logic and uses natural deduction (not the resolution principle). The base of facts consists of triples of objects. A Naur-Backus form of the syntactic description of the base of rules is presented (exact definitions are given by the second author in: ''Un moteur d'inferences pour systèmes-experts en logique du premier ordre: SNARK'', Bull. Liaison Rech. Informatique Automatique 97, 24-28 (1984)). Since different ways of using relations in the rules are possible, SNARK can work in either the logic of propositions or the second-order predicate logic. SNARK allows the management of uncertainty by means of weighted facts and inferences. Some applications in which SNARK has been used are listed.