Case Study: eBay achieves machine-learning-driven conversational commerce with Neo4j

eBay - Customer Case Study

eBay’s New Product Development Group faced the challenge of encoding human decision-making into a knowledge graph to power conversational commerce — specifically its virtual shopping assistant, eBay ShopBot. The team needed a way to combine natural language understanding and machine learning with a graph database to drive realistic, salesperson-like conversations.

They implemented Neo4j as a probabilistic graph model and integrated NLU-driven machine learning directly into the graph, which enabled ShopBot to manage conversations and made it easy to experiment and move models into production. The approach demonstrated how graph structures can capture human knowledge for better decision-making and conversational experiences.

eBay

Ajinkya Kale

Senior Applied Researcher