News Short Text Classification Based on Bert Model and Fusion Model

• DOI: 10.54097/hset.v34i.5482
• Publication Date: 2023-03-22T06:23:37Z
• Authors: Hongyang Cui; Chentao Wang; Yibo Yu
• Keywords: Training set

Text classification task is one of the most fundamental tasks in NLP, and short news text could be basis for many other tasks. In this paper, we applied a fusion model combining Bert TextRNN with some modified details to expect higher accuracy classification. We used THUCNews as dataset which consists two columns numbers. The original was seperated into three parts: training set, validation set test set. Besides, BERT contains pre-training refers use RNN solve problems. trained these models parallel, then optimal obtained through parameter tuning are added fully-connected layer receive final results by weighting efficiency TextRNN. solves problem over-fitting under-fitting single model, helps obtain better generalization performance. experimental show sharp change loss well model. precision, recall-rate F1-score also evaluated paper. much than has gap 1.76%.