Optimization of Web Service Composition Using Factored Markov Decision Process

The service selection problem has gotten many attentions in Web service composition studies. This paper proposes a novel policy based solution to the service selection problem. A composite Web service is modeled as a factored Markov decision process (MDP), and its execution is under the control of an optimal policy, which is a solution of the factored MDP. A travel agent composite Web service shows that the policy based approach is stable and efficient.