A Reinforcement Learning Based Model for Adaptive Service Quality Management in E-Commerce Websites

Providing high-quality service to all users is a difficult and inefficient strategy for e-commerce providers, especially when Web servers experience overload conditions that cause increased response time and request rejections, leading to user frustration and reduced revenue. In an e-commerce system, customer Web sessions have differing values for service providers. These tend to: give preference to customer Web sessions that are likely to bring more profit by providing better service quality. This paper proposes a reinforcement-learning based adaptive e-commerce system model that adapts the service quality level for different Web sessions within the customer’s navigation in order to maximize total profit. The e-commerce system is considered as an electronic supply chain which includes a network of basic e- providers used to supply e-commerce services for end customers. The learner agent noted as e-commerce supply chain manager (ECSCM) agent allocates a service quality level to the customer’s request based on his/her navigation pattern in the e-commerce Website and selects an optimized combination of service providers to respond to the customer’s request. To evaluate the proposed model, a multi agent framework composed of three agent types, the ECSCM agent, customer agent (buyer/browser) and service provider agent, is employed. Experimental results show that the proposed model improves total profits through cost reduction and revenue enhancement simultaneously and encourages customers to purchase from the Website through service quality adaptation.