Title
Searching for Leaders in Complex Networks with a Topological-Signature-Led Genetic Algorithm
Abstract
We model voting dynamics with an Ising system of nodes on a Watts-Strogatz network. A voter's opinion is a node's spin state, and an opinion propagates through links of the networks. The dynamics reveals two classes of voters: the leaders and the followers. Leaders rarely alter their opinions but have strong influence in their neighbors' opinions, whereas followers often change their opinions. As a leader's opinion changes rarely, a leading node will have a spin state with a long flipping period and a long autocorrelation time. We identify a leading node's topological signatures as a high degree and a high neighbor-degree-sum. By fixing these leaders' spins, we can greatly impact the opinion formation process as the change in a system's magnetization shows. In order to locate the optimal set of leaders who can promote a particular opinion, we search for nodes with leaders' topological signatures. We achieve this search with a prototypical genetic algorithm in which the topological signatures of leaders are the fitness indicators. We also verify that the leaders found with the algorithm induce a comparable impact to a system with those found with an exhaustive search. Therefore, our algorithm, which searches for leaders with their topological signatures, may facilitate the promotion of a particular opinion. Our work does not only describe the emergence of leaders but also characterizes the topological signatures of these potential leaders.
Year
DOI
Venue
2019
10.1109/CEC.2019.8790063
2019 IEEE Congress on Evolutionary Computation (CEC)
Keywords
Field
DocType
opinion formation,social network,Ising model,topological signature,genetic algorithm
Topology,Opinion formation,Brute-force search,Voting,Computer science,Complex network,Genetic algorithm,Autocorrelation
Conference
ISBN
Citations 
PageRank 
978-1-7281-2154-3
0
0.34
References 
Authors
9
2
Name
Order
Citations
PageRank
Chun Yin Yip111.05
Kwok Yip Szeto26421.47