Sentiment Analysis of Hotel Reviews based on BERT and XGBoost

Sentiment analysis of hotel reviews is beneficial for both customers and businesses to quickly gather information and make informed decisions. In order to accurately analyze the sentiment of reviews, a method has been proposed that utilizes BERT for feature extraction from review information, where BERT adopts the MEAN feature extraction strategy, and combines it with the XGBoost model for classification. This method was experimented on publicly available datasets Ctrip_htl_2000 and Ctrip_htl_4000. The experimental results show that the accuracy reached 0.9200 and 0.9213 and the F1 score reached 0.9200 and 0.9212. The experimental results demonstrate that this method outperforms most techniques and achieves good performance.