Text classification system of academic papers based on hybrid Bert-BiGRU model

摘要

The text classification is an important research orientation in the fields of information retrieval and data mining, which has extensive applications in the practical work and scientific research and its research on the algorithm is always a hot topic. At present, the study on the long text like academic texts mainly focuses on abstract extraction. Whereas, due to the complicated content of the text format, there is very little classification research according to the text structure. In this study, the academic texts are classified based on Bidirectional Encoder Representations from Transformers and Bidirectional Gated Recurrent Unit (BERT-BiGRU) model.