Using Cuckoo Optimization Algorithm to Identify Communities in Social Networks

Abstract

Social network analysis helps discover communities and interactions between users. A community is a group of users with high communication density in that group. Although many algorithms have been developed to identify communities, most are inefficient in terms of processing time and cost for large-scale social networks. In this research, we present a simple and efficient algorithm for social recognition in social networks that does not require any prior knowledge about the number of network communities. Most existing methods of community recognition examine the structure of the social network graph without considering issues and interactions between users. In the proposed method, in addition to considering the communication topology between users, we also consider the tweets used by them. The proposed system consists of three general steps. In the first step, the similarity between each user pair is calculated on the basis of a hybrid clustering method. Initial assemblages are based on the similarity matrix in the second step. Finally, in the third stage, using the cuckoo optimization algorithm, the initial clusters are combined and the final clusters are created. In the cuckoo algorithm, the performance of each solution is evaluated using the metrics evaluation criterion on Twitter`s social network. The results show better performance of the proposed method in different criteria than CC-GA and MDCL methods.