Community Detection in Complex Networks Using a Novel Nature Inspired Algorithmic Approach Based on Ant Lion Optimizer

Abstract

Most of the existing system in the world network is represented with the help of links and nodes in which nodes represent the systems and the links represent the relationship between the connecting or interrelating nodes. Some of the well-known existing networks are social media and online social networking websites like Facebook, Google+, and Twitter. The network links in different type of domains represent different types of relationships. E.g. human friendship, animal’s physical proximity, interconnectivity of infrastructures, organizational structures, Web hyperlinks and abstract relationships like similarity between data points. The existence of communities shows the structure of the networks existing in nature. Communities, which are also named as modules or clusters, are the groups of relatively connected nodes, and are said to be intrinsic structures of the networks existing in nature. Nodes of the same community or cluster usually share common interesting properties such as a function, purpose and interest. Hence, one of the most crucial problems in network analysis is community detection. Several methods have been proposed which help in detecting communities of the complex networks. Amongst them, the most popular technique is dependent on the optimization of the objective, modularity which is the most widely used function to evaluate the quality parameters of the group structure of networks. Various heuristic algorithms dependent on modularity optimization in detecting communities have been proposed in past few years. The problem of community detection in complex networks has been received an increased amount of interest since the past decade. Community detection is a way to uncover the structure of networks. This is done by grouping the nodes into communities. The grouping is done for the communities in which the interconnection between the nodes is found to be denser than the intra-connection between the communities. In this thesis, a novel nature-inspired algorithmic approach based on Ant Lion Optimizer is proposed which helps in discovering the communities efficiently in large networks. The proposed algorithm is termed as ALOCD for short. The algorithm optimizes modularity function and is able to identify densely connected groups of nodes having sparse interconnections. Experiments on real-world networks show the capability of the method to efficiently detect the network structure. In this, ALOCD algorithm is also compared with Ant Colony Optimization (ACO) algorithm and Enhanced Firefly Algorithm (EFF) for community detection.