MULTI-LABEL SENTIMENT ANALYSIS MODEL FOR THAI LANGUAGE REVIEW OF RESTAURANT

Abstract

Currently, Sentiment Analysis has been extensively studied, particularly in the context of customer reviews. However, most studies focus on Single-label Sentiment Analysis, which evaluates the overall sentiment of a review as a single entity. In contrast, restaurant reviews often comprise multiple aspects, such as food quality, price, service, and ambience. This study aims to develop a Multi-label Sentiment Analysis Model for Thai restaurant reviews using the Wongnai Review Dataset. The model's performance is evaluated through Traditional Machine Learning approaches, including Logistic Regression, Random Forest, and Support Vector Machine (SVM), in conjunction with text representation techniques such as Bag of Words (BoW) and Term Frequency-Inverse Document Frequency (TF-IDF). The experimental results show that the Random Forest model with TF-IDF achieves superior accuracy of 97.36% in classifying both aspects and sentiments. This study addresses the research gap in multi-aspect sentiment analysis and contributes to the advancement of Thai Natural Language Processing (NLP) applications in the future.