A group abnormal behavior detection method based on deep structure learning

Abstract

The invention discloses a group abnormal behavior detection method based on deep structure learning, and belongs to the technical field of computer information, and the method comprises the steps: building a bipartite graph according to the evaluation behavior of a user on a commodity, enabling a source node in the graph to represent a user account, enabling a sink node to represent the commodity,and enabling a directed edge to represent the feedback record of the user on the commodity; Embedding the source node and the sink node into the same Euclidean space at the same time to obtain low-dimensional representations of all the nodes;and clustering the low-dimensional representations of the nodes to obtain abnormal clusters, namely detected abnormal groups and abnormal behaviors thereof.By means of deep network topology structure information, when the number of abnormal groups is not given to serve as a priori condition, the task of multi-abnormal-group detection is completed, and meanwhile the detection accuracy and expansibility are improved.

Description

technical field [0001] The invention relates to the field of computer information technology, in particular to a group abnormal behavior detection method based on deep structure learning. Background technique [0002] With the vigorous development of the Internet, the transaction volume of e-commerce platforms is increasing. Online users' evaluations and ratings for commodities usually have a great influence on potential users, so there are more and more fake reviews on the Internet. Fraudulent groups comment on the target item multiple times in unison on a large scale. On the one hand, it can achieve the purpose of rapidly increasing the influence of the target item. On the other hand, it can reduce the out-degree of a single fraudster by sharing to avoid the risk of being detected. Therefore, the task of detecting fraudulent groups can be transformed into a density block detection task in a bipartite graph, where source nodes represent user accounts, sink nodes represent ...

AI Technical Summary

Problems solved by technology

[0010] 1. Most of the existing methods use attribute information, but the attribute information in the network is easily tampered by fraudsters and difficult to obtain;

[0011] 2. Even when the number of abnormal groups is given as a priori, the existing technology still does not have a good effect in detecting multiple abnormal groups;

[0012] 3. Only shallow topological structure information can be used, and shallow topological structure information is easily targeted by fraudsters to avoid detection

Images