Machine Learning Algorithm Accurately Analyzes Sentiment in English Texts

Since early sentiment analysis methods lacked understanding of context and had difficulty dealing with complex language structures and polysemous words, this paper introduces a sentiment analysis method based on machine learning algorithms, focusing on multi-granular sentiment analysis of English text reviews. This study first introduces the overall framework of the model, which is the Multi-granularity Emotion Analysis Model based on T5 (MEIAT5) based on the T5 model, adding residual connections (Add layer) and layer normalization (Norm layer) to solve the possible degradation problems of the model. Then, this paper describes the model in detail. The T5 model combines input and output order and attention mechanism to improve the efficiency and accuracy of the model in analyzing text. Furthermore, this paper proposes to perform grammatical analysis by following the grammatical structure of the phrase structure tree in a bottom-up manner. Finally, this study collects different datasets of English texts for experiments. The experimental results show that although the T5+Syntax model combined with grammatical analysis shows better processing capabilities in some situations, its effect is not significantly improved compared with the standard T5 model when dealing with mixed emotions and complex emotional expressions. This study verifies the effectiveness of improvements based on grammatical analysis in certain sentiment analysis tasks, but also points out the possible limitations of current models when dealing with multi-granularity sentiment analysis.