Academic network node classification method and system based on time sequence primitive mode migration

The invention relates to an academic network node classification method and system based on time sequence primitive mode migration. The method comprises the following steps: firstly, acquiring data set processing data of five different field subjects, and constructing an academic database; then constructing a time sequence diagram, taking the papers as nodes, and taking edges as feature information required for storing the papers; extracting a sequence diagram node feature matrix and an adjacent matrix as input, and performing model training; performing time sequence diagram model element mode training on the constructed time sequence diagram; fusing the time sequence diagram model training parameters to obtain time sequence diagram model parameters of the target subject; and finally, carrying out sequence diagram node classification and result optimization of the target subject. According to the method, the GRU recurrent neural architecture is used for updating the weight of the graph convolutional network,