An Interactive Service Composition Model Based on Interactive POMDP

Abstract

Sequential decision making techniques have been applied to solve problems of Cloud service selection and composition. However, existing researches rarely focus on studying Cloud service compositions in a partially observable multiagent interactive environment. In this paper, we propose an interactive service composition model I-SerCom that applies the Interactive Partially Observable Markov Decision Process (I-POMDP) to help service buyers make sequential service selection policies. I-SerCom enables buyers make decisions by considering both changes of states of service marketplaces and an evolution of the intention of other involved agents. We use Interactive Influence Diagram (I-DID) to graphically represent the model. The I-DID factorization makes the model description more explicit and simplifies the model solving process. We use a case study of cloud service selection to present the process of I-SerCom, and conclude that I-SerCom is an efficient method for service composition.