Inductive Logic Programming (ILP) is a form of learning from examples that employs clausal logic as a representation formalism (Muggleton & De Raedt, 1994). Its goal is to generalise, relative to a given background theory, given extensional knowledge to obtain an intensional theory. The resulting theory may consist of a Horn clause predicate definition, or a set of clauses (possibly indefinite clauses or denials) describing relations between predicates. This corresponds well to the traditional distinction in deductive databases between view definitions and integrity constraints.
展开▼
