Hybrid optimization algorithm for community and fraud detection in complex networks for high immunity towards link and node failures

Abstract

The complex networks are offering a high resource of heterogeneous data and the proper and efficient analysis discovers the unknown information and relations in networks. Due to the huge number of users and nonfamiliar fraud detection system in complex networks, a lot of online frauds introduce to affects the networks. In this paper, we concentrate on both community and fraud detection to minimize the link and node failures in the complex networks. A hybrid optimization algorithm proposed for community and fraud detection in the complex networks (HCFD-Net). The first contribution is to detect the community based on fruit fly optimization algorithm with differential evolution (FOADE). The second contribution is that the fraud detection is achieved by contingency table terminology with multi-link metrics. The performance of the HCFD-Net is analyzed on different five real-world networks are Zachary's karate club, Bottlenose dolphins', American college football, American political books, and Amazon online purchase network. The simulation result shows that the proposed HCFD-NET perform very efficient than existing algorithms in terms of normalized mutual information (NMI) and network lifetime.