Skip to main content

Inductive Logic Programming 2

Inductive logic programming (ILP) is a research area lying at the intersection of inductive machine learning and logic programming. The general aim of ILP is to develop theories, techniques and applications of inductive learning from observations and background knowledge in a first order logical framework. The ILP2 project is an ESPRIT Long Term Research project which started on January 1, 1996, and lasts for 3 years. It continues the ILP project (September 1992 - August 1995). Further general information about the ILP2 project can be found on the ILP2 home page. The following is a description of the work done at Bristol.

Recent work has concentrated on the realisation of a practical system for descriptive induction of integrity constraints, that can be applied to real-world data mining tasks (Flach, 1997; Flach & Lachiche, 1997). A first release of the Tertius system is now available. Furthermore, we have investigated the use of a strongly typed language as conceptual basis for ILP (Flach, Giraud-Carrier & Lloyd, 1998). This line of research has resulted in the 1BC first-order Bayesian classifier (Flach & Lachiche, 1999). Finally, we have continued previous work on logical characterisation of inductive reasoning (Flach, 1998).

Staff and Students

Peter Flach, Christophe Giraud-Carrier, Nicolas Lachiche, and Torbjorn Dahl.

Publications

Support

This research is supported by the European Union (ESPRIT LTR 20237)