Training method and detection method for generative adversarial multi-relation graph network

Abstract

The invention discloses a training method of a generative adversarial multi-relation graph network model for detecting a machine account, and the training method comprises the steps: modeling a platform into a graph containing a node v and a relation r, and the number of the graph is determined by the type number of the relation r; generating a false target node vt of the source node v by using a generator G; respectively inputting the sampled node pairs (v, u) and (v, vt) into a connection relation discriminator D, and repeatedly training the connection relation discriminator D; reasoning node pairs in the graph by using the trained connection relation discriminator D, determining the connection relation of the node pairs, and further updating the structure of the graph; and inputting the representation vectors of the nodes into a classifier, updating parameters of the model according to a loss function and back propagation, and carrying out multiple times of training to obtain a trained generative adversarial multi-relational graph network model. The invention also discloses a machine account detection method based on the generative adversarial multi-relation graph network model.

Description

technical field [0001] The present application relates to the field of machine account detection, and in particular to a training method and a detection method for generating an adversarial multi-relational graph network model for detecting machine accounts. Background technique [0002] Since 2016, the third generation of machine accounts has been discovered. Due to the deepening of human operation and automation, these accounts even steal information from other real accounts, and use artificial intelligence technology to generate highly credible text or pictures, which behave more like Real human accounts make machine accounts more difficult to detect and identify. [0003] At present, many patents on machine account detection methods have been proposed. For example, distinguish between normal accounts and robot accounts by analyzing the social relationships of users’ friends; use posting and following strategies for honeypot accounts to collect accounts, detect robot acc...

AI Technical Summary

Problems solved by technology

[0002]Since 2016, the third generation of machine accounts has been discovered. Due to the deepening of the mixture of human operation and automation, these accounts even steal information from other real accounts. Intelligent technology generates highly credible text or pictures, which behave more like real human accounts, making machine accounts more difficult to detect and identify

[0004] However, the above existing technologies face the following problems: relying on a large number of data samples, the detection model is not proposed in time; the existing detection scheme uses the collected machine account data for analysis and training, and then proposes a detection model corresponding to the training set and obtains better results. However, the generalization of these methods is poor

Images