Marketing cheating monitoring method based on heterogeneous graph convolutional neural network

The invention discloses a marketing cheating monitoring method based on a heterogeneous graph convolutional neural network, and the method comprises the following steps: S1, constructing a feature vector of a to-be-detected user based on the historical consumption features and behavior sequences of the user, and taking the feature vector as a feature for short; s2, constructing a heterogeneous user association graph based on the login relationship of the user-equipment and the consumption relationship of the user-merchant; s3, training various sub-graphs of the user by using a graph convolutional neural network model to obtain feature vectors of the user based on different graph structures; the method has the advantages that the heterogeneous graph data structure is adopted, compared withan isomorphic graph structure, the relationship types are further enriched, more topological structure information is introduced, the problem that different relation information has different weightsis solved, and good expans