Predicate Invention from a Few Examples

摘要

One main difficulty of Inductive Logic Programming lies in learning recursively defined predicates. Today's systems strongly rely on a set of supporting predicates known as the background knowledge, which assumes that the user knows in advance what sort of predicates are required by the target definition. Predicate invention can remedy the situation by extending the speccification language with new concepts that appear netiher in the examples nor in the background knowledge, and finding a definition for them. A serious concern is that no examples of the invented predicate are explicitly given but rather of the target predicate, so learning has to be done in the absence or scarcity of examples. This work shows an autonomous leanring method based on inverting clausal implication that can ivnent the recursive predicates it needs. The learner is endowed with means to work from a small data set.