Periodical recommendation method based on multi-granularity heterogeneous attribute graph comparative learning

The invention discloses a journal recommendation method based on multi-granularity heterogeneous attribute graph comparative learning, and relates to the technical field of data analysis, and the journal recommendation method comprises the following steps: data processing, multi-granularity semantic feature extraction, multi-granularity structural feature extraction based on heterogeneous graph neural network comparative learning, and adaptive learning. On the basis of a heterogeneous graph neural network, comparative learning, a convolutional neural network and a text preprocessing method, a text is processed into multiple granularities and is respectively processed from semantic and structural directions, and the processed features are integrated into a final optimal text feature by using hierarchical adaptive learning. And softmax is used to obtain a final classification result. And a plurality of loss functions are integrated to train the model in consideration of particularity of papers and periodicals.