Results 1 
2 of
2
A latticebased approach to hierarchical clustering
 Proceedings of the Florida Artificial Intelligence Research Symposium
, 2001
"... Abstract The paper presents an approach to hierarchical clustering based on the use of a least general generalization (lgg) operator to induce a lattice structure of clusters and a category utility objective function to evaluate the clustering quality. The objective function is integrated with a la ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Abstract The paper presents an approach to hierarchical clustering based on the use of a least general generalization (lgg) operator to induce a lattice structure of clusters and a category utility objective function to evaluate the clustering quality. The objective function is integrated with a latticebased distance measure into a bottomup control strategy for clustering. Experiments with wellknown datasets are discussed. Introduction In the context of Machine Learning clustering is an approach to discovering structure in data. Therefore the clusters are usually represented intensionally (in terms of relations, properties, features etc.) and hierarchical clustering techniques are of main interest. The term often used for this area is conceptual clustering. Three basic issues are important in conceptual clustering: the type of cluster representation, the control strategy used to search the space of possible clusterings and the objective f~mction used to evaluate the quality of clustering. Clusters can be represented by necessary and sufficient conditions for cluster membership (e.g. rules, decision trees) or probabilistically by specifying the probability distribution of attribute values for the members of each cluster. The control strategy determines the way in which the clustering tree is generated (e.g. topdown, bottomup, hierarchical sorting). The objective function can be integrated in the control strategy or used separately. The classical conceptual clustering system COBWEB Recently a distancebased approach to creating concept hierarchies has been proposed in
Coveragebased semidistance between Horn clauses
, 2000
"... . In the present paper we use the approach of height functions to defining a semidistance measure between Horn clauses. This appraoch is already discussed elsewhere in the framework of propositional and simple first order languages (atoms). Hereafter we prove its applicability for Horn clauses. ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
. In the present paper we use the approach of height functions to defining a semidistance measure between Horn clauses. This appraoch is already discussed elsewhere in the framework of propositional and simple first order languages (atoms). Hereafter we prove its applicability for Horn clauses. We use some basic results from lattice theory and introduce a family of language independent coveragebased height functions. Then we show how these results apply to Horn clauses. We also show an example of conceptual clustering of first order atoms, where the hypotheses are Horn clauses. 1 Introduction Almost all approaches to inductive learning are based on generalization and/or specialization hierarchies. These hierarchies represent the hypothesis space which in most cases is a partially ordered set under some generality ordering. The properties of partially ordered sets are well studied in lattice theory. One concept from this theory is mostly used in inductive learning  this is...