Abstract
Recommender systems are an integral part of many commercial platforms. A subset of recommended products is the result of the aggregated behavior of users who also purchased those products, known as the user-generated network. User-generated product reviews, images, and hashtags are increasingly valuable sources of information for customers to make product decisions online. A recent work stream addressed the economic impact of the review. Typically, the influence of product reviews is explained by numerical variables representing the value and number of reviews. On the other hand, the platform itself provides some products related to these products in a more generic way, for example: B. Product categories. Product recommendations are basically a filtering system that tries to predict and display items that a user might want to buy. It might not be completely accurate, but if it's showing you something you like, then it's working. This set of recommended products is called a system-generated network. Our goal is to capture the evolution of product networks in terms of link formation and removal as a consequence of any of the above suggestions by taking into account the strength of nodes in the network. Our results show that while user-generated recommendations have a strong impact on link formation and rank distribution in product networks, system-generated networks have no significant effect.