Application and optimization of deep learning in emotional polarity analysis of music review texts

Abstract

With the rapid growth in the number of music platform users, automated analysis of the emotional polarity of music review texts has become a key means to understand user preferences and improve user experience. This study launched a comprehensive discussion and empirical analysis around the application and optimization of deep learning technology in emotional polarity analysis of music review texts. By comparing the performance of models such as Long Short-Term Memory Network (LSTM), Convolutional Neural Network (CNN) and Transformer in processing music review text, this study found that the Transformer model, with its unique attention mechanism, is better at capturing long-distance dependencies and understanding It performs well in expressing complex emotions, thus outperforming other models in both accuracy and F1 score. In addition, this article details the complete process from data preprocessing to model training and evaluation, including key steps such as text cleaning, word segmentation, stop word removal, and word embedding. Through case studies, the effectiveness and practicality of the deep learning model in actual music review sentiment analysis tasks is further confirmed. This research not only provides a new technical perspective for music review sentiment analysis, but also opens up new paths for future research and applications in this field.