Generating Contrastive Explanations for Inductive Logic Programming Based on a Near Miss Approach