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

Providing high-quality service to all users is adifficult and inefficient strategy for e-commerce providers,especially when Web servers experience overload condi-tions that cause increased response time and requestrejections, leading to user frustration and reduced revenue.In an e-commerce system, customer Web sessions havediffering values for service providers. These tend to: givepreference to customer Web sessions that are likely tobring more profit by providing better service quality. Thispaper proposes a reinforcement-learning based adaptivee-commerce system model that adapts the service qualitylevel for different Web sessions within the customer’snavigation in order to maximize total profit. The e-com-merce system is considered as an electronic supply chainwhich includes a network of basic e- providers used tosupply e-commerce services for end customers. The learneragent noted as e-commerce supply chain manager(ECSCM) agent allocates a service quality level to thecustomer’s request based on his/her navigation pattern inthe e-commerce Website and selects an optimized combi-nation of service providers to respond to the customer’srequest. To evaluate the proposed model, a multi agentframework composed of three agent types, the ECSCMagent, customer agent (buyer/browser) and service provideragent, is employed. Experimental results show that theproposed model improves total profits through costreduction and revenue enhancement simultaneously andencourages customers to purchase from the Websitethrough service quality adaptation