An intuitionistic fuzzy data-driven product ranking model using sentiment analysis and multi-criteria decision-making

Abstract

With the advent of web 2.0 websites, the impact of customers on each other's purchasing decisions has increased, and this has changed the decision-making procedure, attitude, and consumer purchase behavior. Along with increasing the volume of customers' reviews on sites, the need for a method that can be used to rank alternative products considering a set of product features and the customer comments related to these features on websites is essential. Many attempts have been made aimed at product ranking through online customer reviews (OCRs), which have their own shortcomings. These limitations occur during different stages, including focusing only on specified features on online shopping sites or extracting features based on term frequency. Other limitations are not paying attention to low repetition, but to only important features (in identifying features), not paying attention to neutral expressions including hesitations or uncertainty in consumers’ purchase decisions (in identifying the sentiment orientations of each review), using expert-based approaches to determine the weight of features (in determining the weight of features), and ranking merely based on the star rating and ignoring the valuable information in OCRs or not considering the robustness of decision-making and most effective features (in ranking alternatives). Accordingly, this study aims to narrow these gaps by proposing an integrated framework that combines sentiment analysis (SA) and multi-criteria decision-making (MCDM) techniques based on intuitionistic fuzzy sets (IFS). In the following, the developed hybrid model has been used in a real-world case to rank five mobile phone products using the OCRs on the Amazon site to illustrate the availability and utility of the proposed methodology. Sensitivity analysis has been performed to determine the most robust method and most effective features.