Research And Application Of Text Classification Based On Improved LSTM Model

In the era of big data,various industries have generated a large amount of data,which can be divided into structured data and unstructured data.For structured data,there are quite a few well-established research methods.But 80% of the data generated by all walks of life is unstructured data,such as text,voice,video,etc.In order to solve the above problems,short text data has problems such as irregularity,sparse context information,and semantic and grammatical interference.This paper introduces text classification technology to enhance users’ cognition and understanding of document objects,thereby helping users to quickly filter out valuable information from massive text data.Taking the review data of a MOOC course as the research object,this paper explores and analyzes the sentiment classification of Chinese short text data.The main research contents are as follows:First,use the web crawler technology to obtain a MOOC course comment dataset and a Taobao store comment dataset that has been manually labeled,and use the BERT pre-training model to label the MOOC course comment dataset with pseudo-labels according to different emotional characteristics.Then,the text classification method based on topic,the text classification method based on shallow learning and the text classification method based on deep learning are compared experimentally.optimal.Finally,based on the LSTM model,the relational network is stacked downstream of its network and the attention mechanism is integrated to improve the LSTM model.The text classification performance of the improved LSTM model is compared with other text classification methods on the dataset of this paper.The experimental results show that: the improved LSTM model proposed in this paper has improved various text classification indicators,and the classification performance is the best.Therefore,the text classification method proposed in this paper has certain practical significance and practical value.