Directed social network false user detection method based on homogeneity prediction

Abstract

The invention discloses a directed social network false user detection method based on homogeneity prediction, and the method comprises the steps: firstly enabling a label of each node in a directed social network to be associated with a binary random variable, and carrying out the modeling of the joint distribution of all variables through a paired Markov random field; the joint distribution is a product of a node potential function and an edge potential function; the side potential function is composed of a bidirectional side potential function and a unidirectional side potential function; then, based on a given training set, estimating posterior probability distribution of the nodes by using LBP so as to perform classification or sorting, thereby detecting false users of the directed social network; and maintaining a pair of correction factors for each benign tail node and each Sybil head node of the edge during LBP iteration. According to the method, the edge potential function adaptively adjusts the edge weight for estimating homogeneity, and a direction sensitive mechanism is incorporated, so that the asymmetric interaction between the followers and the followers can be better captured.

Description

technical field [0001] The invention belongs to the technical field of network security, and in particular relates to a method for detecting false users of a directed social network based on homogeneity prediction. Background technique [0002] While a huge influence in social networks, celebrities have many followers, but not all of those followers are actual human beings on the other side of the screen. According to reports, 9%-15% of active Twitter users are bots. Malicious attackers in social networks create and control such bots or Sybil to conduct spam, phishing scams, referral traffic, or manipulate public opinion, causing a series of security issues and trust crises. [0003] In order to combat this chaos in social networks, a variety of Sybil detection methods have emerged. Among them, feature-based and structure-based methods are mainstream. Signature-based methods use various information about the target user, such as user profile, IP address, and various behav...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the problems that the existing LBP-based method ignores the local homogeneity difference of the edge, cannot characterize the behavior node mode, an...

Images