An Approach for Mining Web Service Composition Patterns from Execution Logs

Abstract

A service-oriented application is composed of several web services to provide complex functionality that a single web service cannot provide. A set of services along with their control flows can be frequently used in multiple applications. Such services form a service composition pattern which is well tested in the numerous adoptions. Reusing service composition patterns in service composition provides an efficient way to improve the quality of new applications. To facilitate the documentation of service composition patterns, we propose an approach to automatically recognize service composition patterns from various applications. We identify service composition patterns by locating a set of associated services commonly used by different applications and recovering the control flows among the set of associated services.