Leveraging Knowledge Graphs for E-commerce Product Recommendations

André Gomes Regino, Rodrigo Oliveira Caus, Victor Hochgreb, Júlio Cesar dos Reis

Abstract: The quantity of data produced by e-commerce sales has significantly increased in recent years. Customers frequently ask online stores questions about products, such as price, warranty, and shipping costs. Improving response times can enhance customer satisfaction and increase sales conversion rates. In this context, suggesting alternative products when the desired product is unavailable is crucial for sales growth. This study presents and evaluates a technique for product recommendation that relies on the product data stored in knowledge graphs (KGs). The KG contains facts derived from natural language questions and answers extracted from an e-commerce platform. We demonstrate our approach through a practical application, using data from online stores processed by GoBots, a leading e-commerce chatbot provider in Latin America. We provide quantitative and qualitative analysis to evaluate the quality of our solution and discuss drawbacks and challenges in its adoption.