ANOMALY DETECTION IN COMPLEX NETWORK USING COMMUNITY DETECTION CONCEPT
Azadeh Oliyaei, Alireza Aliahmadi · International Education and Research Journal · 2021
Discovering anomalous users in the social network is a crucial problem in analyzing network. The previous works focus on a network with just one type of interaction among the entities. However, the relationship among people is highly complex, and users have multiple types of interaction in a social network. On the other hand, users tend to form a community in the social network such that normal users usually have friends who are frends themselves, and anomalous users typically do not follow this rule. In this paper, we consider the detection of anomalous nodes in the multi-layer social network by combing the information in each layer of the network. We propose a pioneering algorithm based on the community detection method and assign the anomaly score to each user and rank them. Experimental result on real dataset shows that the proposed algorithm can recognize anomalous users in the multi-layer social network.