First-Order Induction of Decision Lists (FOIDL) is a new ILP method developed as a result of the failure of existing ILP methods when applied to the learning of past tense of English verbs. Using intentional knowledge representation, output completeness, and first-order decision lists, it can learn highly accurate rules for the past tense problem but with some drawbacks. In this paper, we present an extension to FOIDL that learns rules faster using a learning algorithm similar to version space learning.
Published Date: May 1999
Registration: ISBN 978-1-57735-080-4
Copyright: Published by The AAAI Press, Menlo Park, California.