Sentiment-Based Rating Prediction Using AFINN Lexicon: Insights for Product Promotion in Digital Marketing Analytics

The tourism and hospitality industry significantly impacts the global economy in the big data era, highlighting the role of technological advances, especially in social media, in shaping visitor preferences and behavior. This research focuses on how sentiment analysis is assisted by the Natural Language Processing (NLP) word embedding model and TF-IDF (Term Fre-quency-Inverse Document Frequency) to increase prediction accuracy. The dataset used is tourism data in the hospitality sector. Then it is categorized as positive or negative accurately according to customer preferences, contributing to research in developing smart tourism and improving tourism infrastructure. The evaluation results show that the sentiment analysis prediction model implementing TF-IDF, which was trained using a dataset of 515,212 reviews from users in tourism and hospitality, performs well, reaching an accuracy of 91%. Visualization and sentiment prediction results are carried out using Python.