A text classification network model combining machine learning and deep learning

Text classification is significant in natural language processing tasks, which can deal with a large amount of data scientifically. However, for text feature extraction, it is not easy to simultaneously consider the characteristics of short and long texts. Moreover, it does not reflect the importance of words in the text, resulting in unsatisfactory text classification results. Therefore, this paper proposes a machine learning and deep learning model. Specifically, text features are extracted by joint training, and then an attention mechanism is introduced to classify short texts and long texts. Firstly, the pre-processed data is subjected to term frequency-inverse document frequency, text convolutional neural networks and rotary transformer models for joint extraction of text features. Subsequently, the attention mechanism is introduced for the weight distribution problem after model fusion to improve the focus on keywords. Eventually, the experimental results indicate that the model proposed in this paper has a good effect on long and short-text classification. We achieved 95.8%, 92.5% and 95.4% accuracy on three public datasets, respectively. In this way, the proposed model is significant in text classification.