Abstract
In the past few decades, although people have conducted in-depth research on community detection in one-mode networks, community detection in bipartite networks has not been extensively researched. In this paper, we propose an improved artificial bee colony algorithm named IABC-BN, which is used to detect the communities in two-mode graphs (i.e. bipartite graphs) with two kinds of vertices in the cluster (i.e. community). Firstly, this paper proposed a novel population initialization process of artificial bee colony (ABC) method for two-mode graph cluster identification. This initialization method can improve the diversity of initial population of ABC and speed up its convergence rate. Secondly, in the employed bee search step of the algorithm, a new combinatorial search equation is proposed. This equation is guided by the global optimal solution and the better neighbour solution of the current solution. By using this combination equation and the increased parameter perturbation frequency, the exploitation ability of the algorithm is further enhanced. Thirdly, in the onlooker bees step, another new combination search equation is also proposed. This equation improves the exploitation level of the algorithm, and an opposition based studying method is employed to promote the exploitation ability of the algorithm. Lastly, in scout bee stage, a probability threshold $\beta $ is introduced to enhance the exploration ability of the algorithm and improve the population diversity of the algorithm. To our knowledge, the IABC-BN method presented in this paper is the first ABC method used to cluster identification in two-mode graphs with two kinds of vertices in the cluster. For verifying the accuracy of the results of the proposed method, a large number of experiments are carried out making use of synthetic bipartite graphs and real bipartite graphs. The test outcomes show that this algorithm is an excellent algorithm for cluster discovery in two-mode graph.
Highlights
Network science is a basic subject across computer science, social science, biology and other disciplines
Parameters: number of food sources SN, maximum cycle number MCN, modification rate MR, a threshold value used in the onlooker bees phase α, number of cycles without ameliorating limit, a value β of threshold used in the scout bees phase; Output:a cluster partition C of the graph; 1: pop= The original food source swarm achieved by employing algorithm 1; 2: Initialize triali = 0 (i=1, 2, . . . , SN); 3: cycle=1; 4: while cycle ≤ MCN do
IABC-BN algorithm 38: Calculate the probability value priq base on formula (13); 39: Stochastically select a neighbour with an odds priq; 40: Two integers a, b are produced randomly and a = b = i; 41: Calculate W= MR × n ; 42: W different integers are randomly generated from the set [1..n], and the result is represented by permutation; 43: for w=1 to W do 44: j= permutation (w); 45: r=rand(0,1); //rand (0,1) is a stochastic number //between 0 and 1
Summary
Network science is a basic subject across computer science, social science, biology and other disciplines. Complex graphs stand for different complex systems of various subjects. Complex networks consist of nodes (or vertices) and links (or edges). A vertex stands for an individual of a complex graph, and a link represents the interaction and communication between two individuals of a complex graph. Biological researchers try to comprehend the relevance between disease genes and all known phenotypes from a network of disease genes and disorders. Sociologists study the behavior characteristics of various user groups in online social networks. A lot of other examples can be from transportation, computer science, marketing, economics, politics, etc
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
