In recent years, a broad learning system that attempts to replace deep structure learning methods has been proposed. In text classification, the traditional method LSTM is used in most cases. two methods of breadth learning are used in the paper for the classification task of movie review data, namely augmented nodes and simultaneously augmented and mapped nodes, and the training results are compared with those of LSTM. Previously, broad learning has achieved good results on some light datasets, but the movie review dataset is a larger dataset in comparison, with the passage of time, broad learning seems to fade out of the limelight, and few scholars are seen to use it anymore. In this paper, we will experimentally explore whether broad learning can outperform the traditional method LSTM on movie review datasets, and find out the advantages and disadvantages of bbroad learning, and explore the reasons for the decreasing hotness of breadth learning.