Summary
International Symposium on Nonlinear Theory and its Applications
2017
Session Number:A3L-E-2
Session:
Number:A3L-E-2-5
A Graph Clustering Method Using Genetic Algorithm
Goutam Chakraborty, Naoki Sato,
pp.334-337
Publication Date:2017/12/4
Online ISSN:2188-5079
DOI:10.34385/proc.29.A3L-E-2-5
Summary:
In several domains, the available information can be represented as a graph where an unit of information is a node. Links are relations between two units of information. For example, two web-sites on the world-wide-web are connected, if there is a hyperlink from one to the other. Members of a social network are connected as a graph. Scientific papers can be connected in many ways, through common key-words, references, or authorship. Most of such networks form communities, where within communities the relations (node interconnections) are strong, whereas between communities the links are less. Finding those communities, and finding the important (central) nodes of a community are two main problems, for several mining applications. Identifying communities is Clustering the network. Most of the existing algorithms, like k-means, need a pre-defined value of the number of clusters. In this work, we introduce a genetic algorithm based approach, where the optimum number of clusters are automatically determined, through a few generations of the genetic search.