Research and Implementation of Text Classification Based on BERT Model

This paper focuses on automatic text classification using the BERT model. [1] It optimizes word vectors in the embedding layer, extracts context features with a two - layer GRU, and classifies via the attention mechanism. Model comparisons show that the BERT - two - layer GRU model has high accuracy. Using BERT and adding title - based keywords can enhance classification accuracy.