Novel Trends in Clustering - Evolving Application Domains of Data Warehousing and Mining - page 185

Chapter 9
Novel Trends in Clustering

Claudia Plant
Technische Universität München, Munich Germany, Ludwig Maximilians Universität München,
Munich, Germany

Christian Böhm
Technische Universität München, Munich Germany, Ludwig Maximilians Universität München,
Munich, Germany

ABStrAct

Clustering or finding a natural grouping of a data set is essential for knowledge discovery in many
applications. This chapter provides an overview on emerging trends within the vital research area of
clustering including subspace and projected clustering, correlation clustering, semi-supervised clus-
tering, spectral clustering and parameter-free clustering. To raise the awareness of the reader for the
challenges associated with clustering, the chapter first provides a general problem specification and
introduces basic clustering paradigms. The requirements from concrete example applications in life
sciences and the web provide the motivation for the discussion of novel approaches to clustering. Thus,
this chapter is intended to appeal to all those interested in the state-of-the art in clustering including
basic researchers as well as practitioners.

IntroductIon

to disease onset and progression. Small molecules
involved in primary and intermediate metabo-
lism are called metabolites. Metabolite profiling
provides techniques to quantify the amount of
metabolites in a sample. Due to recent advances of
high-throughput technologies such as tandem mass
spectrometry (MS/MS) hundreds of metabolites
can be detected from a single blood sample. As a
second example consider web usage. For each user
accessing a page, the corresponding web server logs
information including IP address, time of access,
file path, browser and amount of transferred data.

In many applications, for example in medicine, life
sciences, physics and market observation, terabytes
of data is collected every day. Consider for example
metabolite profiling (Baumgartner & Graber 2008).
As an evolving branch of life sciences, Metabolom-
ics studies the highly complex metabolism of cells,
tissues, organs and organisms. One major focus of
research is on identifying subtle changes related