Jiang, J. Q.; Dress, A. W. & Yang, G.: A spectral clustering-based framework for detecting community structures in complex networks. In: Applied Mathematics Letters 22 (2009), Nr. 9, S. 1479 - 1482
[Volltext]
Exploring recent developments in spectral clustering, we discovered that relaxing a spectral reformulation of Newman's Q-measure (a measure that may guide the search for-and help to evaluate the fit of - community structures in networks) yields a new framework for use in detecting fuzzy communities and identifying so-called unstable nodes. In this note, we present and illustrate this approach, which we expect to further enhance our understanding of the intrinsic structure of networks and of network-based clustering procedures. We applied a variation of the fuzzy k-means algorithm, an instance of our framework, to two social networks. The computational results illustrate its potential.
@article{Jiang20091479,
author = {Jiang, Jeffrey Q. and Dress, Andreas W.M. and Yang, Genke},
title = {A spectral clustering-based framework for detecting community structures in complex networks},
journal = {Applied Mathematics Letters},
year = {2009},
volume = {22},
number = {9},
pages = {1479 - 1482},
url = {http://www.sciencedirect.com/science/article/B6TY9-4W6XYH5-5/2/693a9ed19784792496c83e96b4fa828b},
doi = {10.1016/j.aml.2009.02.005},
keywords = {detection, clustering, spectral, community, COMMUNE},
abstract = {Exploring recent developments in spectral clustering, we discovered that relaxing a spectral reformulation of Newman's Q-measure (a measure that may guide the search for-and help to evaluate the fit of - community structures in networks) yields a new framework for use in detecting fuzzy communities and identifying so-called unstable nodes. In this note, we present and illustrate this approach, which we expect to further enhance our understanding of the intrinsic structure of networks and of network-based clustering procedures. We applied a variation of the fuzzy k-means algorithm, an instance of our framework, to two social networks. The computational results illustrate its potential.}
}