Abstract: As a key information source in the modern consumer decision-making process, the depth and breadth of information contained in online reviews have a significant role in guiding consumers’ purchasing behavior. In order to effectively use such information to assist consumers in making reasonable decisions, this paper constructs a probabilistic product ranking model that integrates user consensus reviews and consumer personalized preferences. The model realizes the in-depth excavation of complex review information and the effective utilization of consumers’ personalized preference through five core modules systematically. Firstly, using sentiment analysis technology, star ratings and text reviews are organically integrated to form the user’s individual evaluation domain (IED); subsequently, based on the interval number density operator and interval similarity hierarchical clustering algorithm, etc., the comprehensive evaluation domain (CED) of the commodity’s performance index is solved, which embodies the user’s high degree of consensus; next, with the help of the sequential relationship analysis method to express consumers' individual preferences intuitively, the personalized calculation of the weights of the performance indexes is realized; further, based on Monte Carlo simulation, the stochastic aggregation method is used to solve the superiority matrix reflecting the relative superiority and inferiority relationships among commodities, which lays the data foundation for the probabilistic ranking solution of commodities; finally, an iterative ranking method of superiority based on damped optimization (IRS-DO) to obtain a probabilistic ranking conclusion set that both reflects individual consumer preferences and has high robustness. Based on the theoretical method, a validation analysis is carried out with a real case of new energy vehicle, which clarifies the characteristics and application effect of the ranking model constructed in this study. This study further enriches the online review information-driven commodity recommendation ranking method system, which has a good application prospect in the field of personalized recommendation of commodities and customized service of platforms.

Key words: Online reviews, IDWAWAA operator, Stochastic aggregation, Information fusion, Decision support