In the line of previous work by S. Muggleton and C. Sakama, we extend the logical characterization of inductive logic programming, to normal logic programs under the stable models semantics. A logic program in this non-monotonic semantics can be contradictory or can have one or several models. We provide a complete characterization on the hypotheses solution to induction of this kind of programs. 							
						
							
					 															
					Ramón P. Otero