This study utilizes natural language processing (NLP) and deep learning techniques to analyze and interpret the Singapore Airlines customer review data set, aiming to extract useful information through advanced text analysis methods to gain a deeper understanding of passenger experience and expectations. The research first constructs the input features of the model through preprocessing, word embedding of review titles and text content. The topic modeling technology is used to identify key topics in the comments, and then the emotional tendency of the comments is accurately predicted through the recurrent neural network (RNN) and long short-term memory network (LSTM) in the sequence model. Research results show that the LSTM model has better performance and better gradient stability than RNN in processing long text data. It can effectively capture long-distance semantic dependencies and improve the accuracy of emotion classification. In addition, topic modeling revealed multiple themes in passenger reviews, such as service quality, flight experience, etc., providing Singapore Airlines with operational business insights to help the airline optimize services and improve customer satisfaction.