Symbolic clustering

Clustering is the problem of finding a good organization for data. Because there are many kinds of clustering problems, and because there are many possible clusterings for any data set, clustering programs use knowledge and assumptions about individual problems to make clustering tractable. Cluster analysis techniques allow knowledge to be expressed in the choice of a pairwise distance measure and in the choice of clustering algorithm. Conceptual clustering adds knowledge and preferences about cluster descriptions. In this dissertation, I describe symbolic clustering, which adds representation choice to the set of ways a data analyst can use problem-specific knowledge. I will develop an informal model for symbolic clustering, and use it to suggest where and how knowledge can be expressed in clustering. A language for creating symbolic clusterers, based on the model, has been developed and tested on three real clustering problems. The dissertation concludes with a discussion of the implications of the model and the results for clustering in general.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2011

Author dc:creator

• Reinke, Robert Eugene

Contributors dc:contributor

• Baskin, Arthur B., III

• Artificial Intelligence
• Computer Science

Statement dc:rights

• Copyright 1991 Reinke, Robert Eugene

Language dc:language

eng

Identifier

AAI9210964
(UMI)AAI9210964

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/19061