リンクの向きに着目した機能コミュニティとモチーフの関係分析

卓恭 伏見, 和巳 斉藤, 哲夫 池田 et al. · Institutional Repositories DataBase (IRDB) · 2013

In this paper, in order to detect nodes' functions such as sending/receiving information, one-way/bidirectional relationships and so forth, we propose a method for extracting communities each of which consists of functionally similar nodes from directed networks. We confirm effectiveness and usefulness of our proposed method in comparison with two methods, a standard functional community extraction method intended for undirected networks and a method based on network motif analysis which reveals local link structures. From our experimental results using artificial and real networks, we show that our method can extract some reasonable functional communities which can not be extracted by two comparison methods. We also analyze the values of global jump probabilities which affect the results of community extraction in the PageRank calculation step of our proposed method. We show that, when we set the values of global jump probabilities as α≃0, then, we can obtain reasonable communities by various function of nodes from our experiments.

Cited by 0 works.

Authors