Twitter-Based E-Commerce Recommender System Using a Directed Multilayer Network

Abstract-An e-commerce recommender system seeks to improve the shopping experience and increase sales by providingpersonalized product suggestions based on users’ preferences, behaviors, and interactions. This study combines socialnetwork analysis with a directed multilayer network framework to enhance product recommendation using Twitter data.User interactions are represented as a weighted directed network to identify influential users and capture the structure ofsocial relationships, while the multilayer model includes a user layer and a conversation or topic layer connected throughusers' participation in discussions. By jointly analyzing network topology, purchase history, shared interests, and socialinteractions, the framework recommends products to users and their social connections. The use of self-reported Twitterinformation further supports socially aware and personalized recommendations, demonstrating the effectiveness of directedmultilayer networks for improving recommendation quality and enabling targeted product promotion in onlinecommunities.