Detecting abusive collaborative activity using a graph neural network

Abstract

A technique uses a graph neural network (GNN) to determine whether a particular entity under consideration is engaging in abusive network-related activity over a computing network in collaboration with other entities. In some applications, the particular entity is part of a bot attack aimed at fraudulently engaging with advertisements. The technique trains the GNN by performing machine learning on a training set that includes a plurality of nodes, edges, and node labels. In forming the training set, the technique associates a feature set with each node in the training set that describes the network activity exhibited by that node's entity. The technique then connects each pair of nodes in the training set with an edge if the feature sets of the pair satisfy a prescribed test for similarity. The technique assigns labels to at least some nodes to convey whether the nodes are associated abusive network-related activity.