Recommendation systems and convergence of online reviews: The type of product network matters!

Abstract

This paper examines the association between product networks generated by recommendation systems and the product ratings' convergence of products. It further investigates how different types of product networks are associated with a customers' perception of quality between product pairs in a product network. Additionally, this study examines whether the type of product networks are associated with the convergence of expressed sentiments for products in a product network differently. Data is collected from a major Swiss e-commerce platform and analyzed via various econometrics models. The results suggest that a network connection between two products via a “substitute” recommender system leads to a convergence of the two products' ratings. In contrast, a connection via a “complementary” recommender system leads to the product ratings' divergence. Additionally, consistent findings are shown to be present when examining the association between the network connection and product sentiment convergence. The findings contribute to WOM and recommendation systems literature and provide interesting implications of the practice.