Integrating Abduction and Induction

We propose an approach for the integration of abduction and induction in the context of Logic Programming. The integration is obtained by extending an Inductive Logic Programming system with abductive reasoning capabilities. In the resulting system, abduction is used to make assumptions in order to cover positive examples and avoid the coverage of negative ones. The assumptions generated can then be generalized in their turn. The system provides a framework for solving tasks that are diicult or impossible for most Inductive Logic Programming systems, such as: learning from incomplete knowledge, learning abductive theories, learning exceptions, learning multiple predicates and learning normal logic programs.