A study into text sentiment analysis model based on deep learning

Deep learning models for text sentiment analysis are employed to analyse the human emotions conveyed by natural language representation in text data. The BERT-ABiLSTM model, a large-scale pre-trained model, is utilised for text data by transforming text word embeddings and extracting global features to analyse the emotions expressed in the text. However, due to the emphasis of ABiLSTM on global features, there are limitations in extracting local features from data. To address this limitation, the TextCNN model is introduced to enhance the local feature extraction capabilities of the model, optimising the process of extracting features from text data. This paper aims to study the text-sensitive analysis model based on the deep learning BERT-CNN-AbiLSTM model. This paper first introduces BERT-AbiLSTM, improves the TextCNN module to enhance local feature extraction, constructs a BERT-CNN-AbiLSTM model, and then analyses the feasibility and reliability of the optimised model through comparative experiments. In the end, this paper introduces the application scenarios of the BERT-CNN-AbiLSTM model to achieve sentiment analysis of natural text language. Copyright © The Authors(s) 2024. Published by Inderscience Publishers Ltd.