Sentiment analysis of hotel comments based on LSTM and GRU

Abstract

Sentiment analysis, which tries to examine the emotional information in the provided text data, has always been a popular topic in the community of natural language processing. Sentiment analysis is currently used in many different contexts, including e-commerce platforms, social media platforms, public opinion platforms, and chatbots. These applications are crucial to the advancement of society and the domestic economy. However, due to the personalization of text data, especially comments, and the presence of acronyms, it is a challenging problem to obtain accurate sentiment information from large and complex unstructured text data. This study presents a comparative examination of various text sentiment analysis approaches, including LSTM, CNN, and GRU. These methods are employed to evaluate their respective performance on sentiment analysis tasks, specifically using a dataset of hotel reviews for training the models. The method presented in this research has been extensively validated through numerous experimental results, affirming its efficacy and its potential to offer novel perspectives for the practical implementation of sentiment analysis.