Web Services Composition with Incomplete QoS Information

Web services composition becomes an ideal choice for enterprise level integration. This paper focuses on the problem of how to select the existed Web services so as to compose a complex service with incomplete QoS information constrains. We first proposed a model of composite service based on Markov process, and then a novel service composition algorithm was designed and implemented, which is built on the k-armed bandit theory to solve the service selection problem. Finally, we carried out extensive experiment on E-Wsframe to verify the solution proposed in the paper. Experimental results show that our method can ensure both the functions and QoS of composite services when selecting and composing Web services at runtime.