Xingli Wu

While the development of e-commerce brings convenience to consumers, a large quantity of products and information increase the difficulty of making purchase decisions. This study constructs a classification-based product selection method driven by online reviews to assist consumers in making purchase decisions. First, the multifaceted attribute evaluations of products are extracted from textual reviews that contain more abundant and useful information than those provided by vendors. The evaluations are modeled by probabilistic linguistic term sets such that sentiment words in texts are described at different frequencies. Then, a classification-based product selection method is developed to rank products considering multifaceted attributes in which alternative products are divided into the acceptance class, rejection class, and uncertainty class through a classification strategy. Each class of products is compared based on the performance scores calculated by a probabilistic linguistic aggregation operator. A case study of selecting laptops based on real data from Amazon.com is given to illustrate the method. Comparative analysis with existing ranking methods shows the advantages of the proposed method in matching consumers’ risk aversion behavior and preserving uncertain information.