9 - A multiobjective Cuckoo Search Algorithm for community detection in social networks

Abstract

This chapter proposes a new model based on the Multiobjective Cuckoo Search Algorithm (MOCSA) for community detection (CD) on social media. The MOCSA model uses a new strategy based on close neighbors' detection in the objective function to increase the CD's accuracy and speed on social networks. The evaluation will be performed on eight datasets such as Karate, Dolphins, Polbooks, Football, Email, Geom, NetScience, and Power Grid. The results show that the normalized mutual information (NMI) value for the Karate, Dolphin, Football, and Polbooks datasets is 1.0000, 0.9984, 0.9486, and 0.7455, respectively. The modularity value for the Karate, Dolphin, Football, and Polbooks datasets is 0.4192, 0.5262, 0.6025, and 0.5264, respectively. The modularity for the Email, Geom, NetScience, and Power datasets are 0.5362, 0.7025, 0.9497, and 0.8382, respectively. Comparisons show that MOCSA performance is better than MOPSO, MOGA, MOMFO, MOFA, MOFPA, MOIWO, MOABC models.