This paper demonstrates how A-Prolog can be used to solve the problem of non-monotonic inductive learning in the context of the learning of the behavior of dynamic domains. Non-monotonic inductive learning is an extension of traditional inductive learning, characterized by the use of default negation in the background knowledge and/or in the clauses being learned. The importance of non-monotonic inductive learning lies in the fact that it allows to learn theories containing defaults and ultimately to help automate the complex task of compiling commonsense knowledge bases.
Subjects: 3.3 Nonmonotonic Reasoning; 12. Machine Learning and Discovery
Submitted: Jan 25, 2007