A network clustering algorithm based on the directionality characterization of network dissimilarity

• Published in: Journal of physics. Conference series Vol. 2898; no. 1; pp. 12034 - 12041
• Main Authors: Jiang, Yuanxiang, Ding, Nan, Shen, Yanli, Zou, Yanni, Li, Meng
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.11.2024
• Subjects: Algorithms; Clustering; Networks; Topology
• ISSN: 1742-6588; 1742-6596; 1742-6596
• DOI: 10.1088/1742-6596/2898/1/012034

Each network or a class of networks often presents specific topological structures. These features describe the relationship between nodes or edges, which directly affects the functionality of the network. Network clustering, as one of the most classical analyses of networks, is highly dependent on the characterization of topological structures and the quantification of network dissimilarity. Given that the existing methods describing the network dissimilarities only consider the difference values but ignore the direction, this paper first proposes a network characterization method based on P-vector for revealing the directionality of network distance. Further, we propose a layout algorithm for multiple networks in a two-dimensional plane and put forward a new network clustering algorithm. The experiments and results indicate the availability and effectiveness of the new proposed methods.