Inverse entailment and Progol-英文文献.pdf

Inverse entailment and Progol Stephen MUGGLETON Oxford University Computing Laboratory Wolfson Building Parks Road Oxford OX QD United Kingdom Abstract This pap er rstly provides a reappraisal of the development of techniques for inverting deduction secondly intro duces Mo deDirected Inverse Entailment MDIE as a generalisation and enhancement of previous approaches and thirdly describ es an implementation of MDIE in the Progol system Progol is implemented in C and available by anonymous ftp The reassessment of previous techniques in terms of inverse entailment leads to new results for learning from p ositive data and inverting implication b etween pairs of clauses Keywords Learning logic programming induction predicate invention inverse resolution inverse entailment information compression Intro duction Since its inception in this journal Inductive Logic Programming ILP has grown to b ecome a substantial subarea of b oth Machine Learning and Logic Pro gramming see The success of the sub ject lies partly in the choice of the core representation language of logic programs Least Herbrand mo dels of logic programs t neatly with the distinction b etween examples and conjectured theories in inductive inference The syntax of logic programs provides mo du lar blo cks which when added or removed generalise or sp ecialise the program Depthb ounded Prolog interpreters used for theoremproving allow ecient test ing of hyp othesised Horn clause theories Most imp ortantly Turingequivalence of logic programs is allowing a broader range of Machine Learning applications in ILP than was p ossible