Inductive logic programming (ILP) studies the learning of (Prolog) logic
programs and other relational knowledge from examples. Most machine learning
algorithms are restricted to finite, propositional, feature-based
representations of examples and concepts and cannot learn complex relational
and recursive knowledge. ILP allows learning with much richer representations.
Our work has focussed on applications of ILP to various problems in
natural language and
theory
refinement for logic programs.