Conceptual clustering, categorization, and polymorphy

Abstract

In this paper we describe WITT, a computational model of categorization and conceptual clustering that has been motivated and guided by research on human categorization. Properties of categories to which humans are sensitive include best or prototypieal members, relative contrasts between categories, and polymorphy (neither necessary nor sufficient feature rules). The system uses pairwise feature correlations to determine the "similarity" between objects and clusters of objects, allowing the system a flexible representation scheme that can model common-feature categories and polymorphous categories. This intercorrelation measure is cast in terms of an information-theoretic evaluation function that directs WITT'S search through the space of clusterings. This information-theoretic similarity metric also can be used to explain basic~level and typicality effects that occur in humans. WITT has been tested on both artificial domains and on data from the 1985 World Almanac, and we have examined the effect of various system parameters on the quality of the model's behavior.

A lattice conceptual clustering system and its application to browsing retrieval

“…Several incremental hierarchical conceptual clustering systems have been developed recently (Lebowitz, 1986;Fisher, 1987;Hanson & Bauer, 1989;Hadzikadic & Yun, 1989;McKusick & Langley, 1991;Martin & Billman, 1994). Although they differ in many respects (e.g., concept representation, hierarchy evaluation criteria, disjoint versus non-disjoint classes), they usually attack the problem by defining a set of hierarchy-change operators and carrying out a hillclimbing search through the space of possible concept hierarchies aimed at finding a "good" hierarchy (Gennari, Langley, & Fisher, 1989).…”

Data mining with graphical models

“…• fuzzy cluster analysis [Bezdek and Pal 1992, Bezdek et al 1999, Höppner et al 1999 segmentation, clustering • conceptual clustering [Michalski and Stepp 1983, Fisher 1987, Hanson and Bauer 1989 segmentation, concept description • and many more Although for each data mining task there are several reliable methods to solve it, there is, as already indicated above, no single method that can solve all tasks. Most methods are tailored to solve a specific task and each of them exhibits different strengths and weaknesses.…”

Untitled

“…Smith & Medin, 1981), or, more generally, as if they possess a feature "polymorphy" (cf. Hanson & Bauer, 1989). Much of the natural world promotes this view: cups, chairs, birds and so forth are labeled as such because they possess smaller feature variance within each category than between categories.…”