Deep Dive into Hotel Reviews Analysis Using Convolutional Neural Networks Algorithm

Tourism industry has been instrumental in the growth of hotels globally, and the rise of online booking systems has led to a surge in online hotel reviews. Currently, travelers are keen to share their experiences, making online hotel reviews a valuable source of information. However, reading and understanding these reviews can be time-consuming and challenging, as human language is complex, and sentiment analysis requires high accuracy. This paper focuses on the development and evaluation of a sentiment analysis model for hotel reviews using Convolutional Neural Networks (CNN). The objective of the model is to assist hotel staff in gaining insights from online reviews. Traditional sentiment analysis algorithms have limitations and have reached a plateau in their performance. The experiment was conducted on six hotel reviews in Langkawi using Word2vec in Python generates word embeddings, and vector representations of words that enable effective natural language processing. The parameters used in the CNN algorithm, such as batch size, filter size, dropout, and data split between training and testing, were varied that need to be experimented with to find the high accuracy of CNN model. The results of the experiments showed that the CNN model achieved high accuracy of more than 90%, making it an effective text classification method. Moreover, the models with a dropout rate of 0.1 consistently outperformed the models with a higher dropout rate of 0.5 in terms of accuracy. The study used the CNN model to create a prototype for predicting the sentiment of hotel reviews. Results showed that the CNN algorithm is a reliable tool for sentiment analysis in the hotel industry, as it provides high accuracy and valuable insights from online reviews.