A Scalable Web Service Composition based on a Strategy Reused Reinforcement Learning Approach

摘要

A central problem in Web services domain is how to get optimal composition of Web services in an uncertain environment. Thousands of Web services published in the internet every day, a large portion of these services may become invalid, deleted or modified. Presently, the environment of Web services changes frequently. In this uncertain service environment, our main object is to find a way to get composite services with good quality of service ( QoS ). A reinforcement learning (RL) approach Q-learning algorithm with strategy reused is presented for Web services selection and composition.