User classification model generation method, device, equipment and computer readable medium

Abstract

The embodiments of the present disclosure disclose a method, device, electronic device and computer-readable medium for generating a user classification model. A specific implementation of the method includes: obtaining a labeled sample set and an unlabeled user behavior feature vector set; for each labeled sample in the labeled sample set, determining from the unlabeled user behavior feature vector set at least one unlabeled user behavior feature vector whose similarity with the sample user behavior feature vector of the labeled sample meets a preset condition; using each sample user behavior feature vector and each determined unlabeled user behavior feature vector as a vertex, and using the similarity of two vertices as the weight of the edge connecting the two vertices, to construct a user relationship network; converting the user relationship network into a graph embedding matrix, and using the graph embedding matrix to train and obtain a user classification model. This implementation is related to artificial intelligence and achieves improved model classification accuracy.

Images