ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

A graph spectral-based scoring scheme for network comparison

Gadiyaram, Vasundhara and Ghosh, Sambit and Vishveshwara, Saraswathi (2017) A graph spectral-based scoring scheme for network comparison. In: Journal of Complex Networks, 5 (2). pp. 219-244. ISSN 2051-1310

Full text not available from this repository.
Official URL: https://doi.org/10.1093/comnet%2Fcnw016


Study of real-world networks is paramount to understand the complex nature of interactions in nature. With a rise in the number of network-based approaches and data availability in various disciplines such as biology, physico-chemical sciences, earth sciences, engineering, economics and social sciences, there is a need for new approaches that are able to capture maximum network features without significant loss of information. The method developed in this article aims to address such requirements and can be applied to study different kinds of networks. Given accurate data points and their connections, it is a challenge to characterize and recognize global patterns. The graph/network spectra are known to hold all the information about the system. A careful analysis of graph spectra should provide the information at any desired level. Here,wehave developed a scoring method to characterize and compare network patterns, using the solutions to normalized Laplacian of weighted matrices. The power of this method is demonstrated on real-world networks by comparing protein structure networks, which is an important problem in structural biology and financial stock networks to study dynamic changes. Various components of the scores provide insights at levels ranging from differences in edges to the transmission of these differences to global clustering levels. The scores developed here not only recognize network patterns in comparison with templates but also can serve as clustering technique to group them in a large pool of networks. Further, given accurate data of nodes and edges in a network, the method is applicable to problems in any discipline of interest.

Item Type: Journal Article
Publication: Journal of Complex Networks
Publisher: Oxford University Press
Additional Information: The copyright of this article belongs to the Oxford University Press.
Keywords: Biological and molecular networks; Network comparison; Spectra of networks; Stock networks; Structural analysis of networks; Weighted networks
Department/Centre: Division of Biological Sciences > Molecular Biophysics Unit
Division of Physical & Mathematical Sciences > Mathematics
Date Deposited: 02 Jun 2022 06:22
Last Modified: 02 Jun 2022 06:22
URI: https://eprints.iisc.ac.in/id/eprint/73058

Actions (login required)

View Item View Item