INTEGRATION OF TEXT CLASSIFICATION AND FUZZY DEMATEL: AN APPLICATION TO E-COMMERCE USER REVIEWS

This study aims to hybrid two methods which are text classification and fuzzy DEMATEL method to find criteria that influence customers to continue buying on ecommerce websites. Text classification is capable of understanding, analyzing, and organizing unstructured data in a short period and capable of providing accurate decisions. Fuzzy DEMATEL helps to overcomes ambiguity or vagueness in decision making. This paper analyses online reviews from a group of e-commerce reviewers based on the products most searched by users for the five most popular e-commerce websites in Malaysia. It is also based on six performance criteria for e-commerce which are, Criteria A (On-time delivery), Criteria B (Customer loyalty), Criteria C (Ease refund), Criteria D (Discount), Criteria E (Product issues), and Criteria F (Order issues). The results of this study indicate that criteria A (On-time delivery) is the most important criterion and Criteria E (Product issues) are the most influencing criteria. This study highlights user's comment using text classification and fuzzy DEMATEL. It is believed to have the more uncertain and subjective result .