Key Laboratory of Ministry of Education, Tianjin Key Laboratory of Intelligence Computing and Novel Software Technology, Tianjin University of Technology, Tianjin, China.

---

Abstract—With continuous emergence of social network platforms, the research of complex network has become a hot field. Complex networks have an obvious feature of community structure, which could be used to study other network characteristics. However, how to better research community structure also becomes a problem that scholars have been exploring. Detecting community structure contributes to analyzing networks to futher discover its implicit patterns. This paper proposes an community discovery algorithm that combines foraging model of ant colony algorithm and signal transmission mechanism to detect overlapping communities. Ants will release pheromones to guide other partners to find the optimal solution, meanwhile pheromones will evaporate at a certain probability. On the other hand, some signals will be lost during transmission. We apply the mechanism of signal loss to process of pheromone evaporation, and consider the similarity between ants to construct ant transfer matrix. Through above two aspects, ant colonies will choose a better walking strategy. In this way, our algorithm can get better division results by adopting above strategy. What’s more, our experiment results indicate that our proposed algorithm could obtain a higher modular value Qov and NMI (Normalied Mutual Information) value, which shows very excellent performance in discovering overlapping communities.

Index Terms—Complex network, community detection, ant colony algorithm, signal transfer, walking strategy.

[PDF]

Cite: Wenjie Li, Xiaoming Yu, "Community Discovery Algorithm Based on Ant Colony and Signal Transfer," Journal of Computers vol. 13, no. 8, pp. 897-904, 2018.