Work

Centralities Based Analysis of Complex Networks

Year: 2012
Type: book-chapter
Abstract: Characterizing, describing, and extracting information from a network is by now one of the main goals of science, since the study of network currently draws the attention of several fields of research, as biology, economics, social science, computer science and so on. The main goal is to analyze networks in order to extract their emergent properties (Bhalla & Iyengar (1999)) and to understand functionality of such complex systems. Two possible analysis approaches can be applied to a complex network: the first based on the study of its topological structure, the second based on the dynamic properties of the system described by the network itself. Since “always structure affects function” (Strogatz (2001)), the topological approach wants to understand networks functionality through the analysis of their structure. For instance, the topological structure of the road network affects critical traffic jam areas, the topology of social networks affects the spread of information and disease and the topology of the power grid affects the robustness and stability of power transmission. Remarkable results have been reached in the topological analysis of networks, concerning the study and characterization of networks structure, and even if far from being complete, several key notions have been introduced. These unifying principles underly the topology of networks belonging to different fields of science. Fundamental are the notions of scale-free network (Barabasi & Albert (1999); Jeong et al. (2000)), cluster (Newman (2006)), network motifs (Milo et al. (2002); Shen-Orr et al. (2002)), small-world property (Watts & Strogatz (1998); Watts (1999); Wagner & Fell (2001)) and centralities. Particularly, centralities have been initially applied to the field of social science (Freeman (1977)) and then to biological networks (Wuchty & Stadler (2003)). Usually, works regarding biological networks rightly consider global properties of the network and when centralities are used, they are often considered from a global point of view, as for example analyzing degree or centralities distribution (Jeong et al. (2000); Wagner & Fell (2001); Wuchty & Stadler (2003); Yamada & Bork (2009); Joy et al. (2005)). A node-oriented approach have been used analyzing attack tolerance of network, where consequences of central nodes deletion are studied (Albert et al. (2000); Crucitti et al. (2004)). But also in this case the analysis have been concentrate on global properties of the network and not on the relevance of the single nodes in the network. Similarly, available software for network analysis is usually oriented to global analysis and characterization of the whole networks. To identify relevant nodes of a biological network, protocols of analysis integrating centralities analysis and lab experimental data are needed and the same for software allowing this kind of analysis. Cytoscape is an excellent visualization and analysis tool with the analysis features greatly enhanced by plug-ins. Plug-in such as NetworkAnalyzer (Assenov et al. (2008)) computes some node centralities but does not allow direct integration with experimental data. Applications such as VisANT 16
Authors: Giovanni Scardoni, Carlo Lau
Language: English
Open Access status: hybrid
OpenAlex ID: W1517881847
FWCI: 12.92
Cites: 43
Cited by: 120
Related to: 10