Abstract
The paper describes a novel computational tool for multiple concept
learning. Unlike previous approaches like ID3 or C4.5, whose major
goal is prediction on unseen instances rathe r than the legibility of
the output, our MPD (Maximally Parsimonious Discrimination) program e
mphasizes the conciseness and intelligibility of the resultant class
descriptions, using three intuitive simplicity criteria to this
end. We illustrate MPD with some additional application s than those
commonly associated with the mentioned algorithms, such as learning
verbal case f rames, translational correspondences or morphological
rules. These include componential analys is (in lexicology and
phonology), language typology, and speech pathology.
full paper