Discussion:

The relevance of a first-class learner for the problem of the past tense and the problems of previous ILP systems in the management of functional tasks whose best representation is the rules with exceptions. Our results clearly demonstrate that an ILP system outperforms both the decision tree and the neural network systems previously applied to the task of the past [4]. This is important because there have been very few results showing that a first-class learner works much better than applying propositional learners to the best coding based on the characteristics of a problem. The FOIL LUCS-KDD implementation described here was used by the LUCS-KDD research group to contrast and compare a variety of CARM algorithms and techniques. The software is available for free, but the author would appreciate proper recognition [5]. The following reference format for referring to the software is suggested.