Progol
Progol is a machine learning system published in 1995 by Stephen Muggleton . It belongs to the paradigm of inductive logical programming and learns definitions of concepts in predicate logic from a set of examples and background knowledge. The examples and background knowledge can be any definite Horn clauses .
Progol combines the bottom-up technique of inverse entailment with a top-down search through the set of clauses . For this purpose, it first constructs the most specific clause for an example, which explains the example together with the background knowledge, in a language restricted by mode declarations . A clause is then determined which subsumes the most specific clause and maximizes a compression metric. This is added to the theory.
In contrast to earlier methods such as FOIL , Progol ensures that the clause it finds is optimal with regard to the metric, since it uses a variant of the A * algorithm.
literature
- Stephen Muggleton: Inverse entailment and progol . New Generation Computing Journal, 13: 245-286, 1995.
- Koichi Furukawa and others: On a sufficient condition for the existence of most specific hypothesis in Progol. In: Nada Lavrač and Sao Deroski (eds.): Inductive Logic Programming. Proceedings of ILP 97. page 157. Springer, 1997.