Authors:
Lars Schütz
1
;
2
;
Korinna Bade
1
and
Andreas Nürnberger
2
Affiliations:
1
Department of Computer Science and Languages, Anhalt University of Applied Sciences, Köthen (Anhalt), Germany
;
2
Faculty of Computer Science, Otto von Guericke University Magdeburg, Magdeburg, Germany
Keyword(s):
Clustering Difference Model, Cluster Comparison, Applications in Planning, Decision Processes.
Abstract:
Clustering data is a major task in machine learning. From a user’s perspective, one particular challenge in this area is the differentiation of at least two clusterings. This is especially true when users have to compare clusterings down to the smallest detail. In this paper, we focus on the identification of such clustering differences. We propose a novel clustering difference model for partitional clusterings. It allows the computational detection of differences between partitional clusterings by keeping a full description of changes in input, output, and model parameters. For this purpose, we also introduce a complete and flexible partitional clustering representation. Both the partitional clustering representation and the partitional clustering difference model can be applied to unsupervised and semi-supervised learning scenarios. Finally, we demonstrate the usefulness of the proposed partitional clustering difference model through its application to real-world use cases in plann
ing and decision processes of the e-participation domain.
(More)