Building a restaurant comment classification model based on machine learning methods

Abstract

Machine learning has been recognized as an efficient classification method, in which the effectiveness of the classification significantly depends on the characteristics of the object extracted from the dataset. With the current explosion of data in general, and especially comment data on online review websites in particular, classification faces numerous challenges. However, accurate classification will have significant implications for consulting activities. This study aimed to build a classification model based on machine learning for restaurant comments data. Data is collected from online review websites such as Tripadvisor.com.vn and Foody.vn. A technique for preprocessing comments to enhance semantic understanding is also proposed. Experiments were conducted using four machine learning algorithms: Naive Bayes (NB), Support Vector Machines (SVM), Decision Tree (DT), and K-Nearest Neighbor (KNN). The results demonstrate that SVM achieves the best classification outcome in comparison to NB, DT, and KNN.