Learning Logic Programs by Discovering Where Not to Search

The goal of inductive logic programming (ILP) is to search for a hypothesis that generalises training examples and background knowledge (BK). To improve performance, we introduce an approach that, before searching hypothesis, first discovers "where not search". We use given BK discover constraints on hypotheses, such as number cannot be both even odd. the bootstrap constraint-driven ILP system. Our experiments multiple domains (including program synthesis general game playing) show our can (i) substantially reduce learning times by up 97%, (ii) scale with millions facts.