Due to the rapid development and widespread use of Internet technology, the Internet has provided a convenient platform for individuals to express their emotions and view comments on microblogs. The collection of user feedback has resulted in a vast amount of data resources, including information on public sentiment. Such data is crucial for the platform's operation and direction, making sentiment analysis of these data sources important for social development. In this paper, we utilize various methods, such as support vector machine (SVM) and recurrent neural network (RNN), including Bayesian, CNN, LSTM, BERT, among others, to analyze the sentiment of online commentary text represented quantitatively as vectors. We evaluate the performance of each model based on accuracy, recall, and F1 score. Our experimental results demonstrate that Bert+CNN outperforms other models in terms of accuracy, recall, and F1 verification. Additionally, we find that LSTM achieves the highest recall, while GRU exhibits the highest F1 score.