Files
Abstract
Web services are autonomous, reusable, platform-independent software applications that can be accessed over the Web. One of the key advantages of utilizing Web services is that they may be quickly composed to form larger processes of varying complexity to provide functionality that none of the individual component services could provide alone. Consequently, organizations have increasingly adopted the use of Web service compositions to assemble complex, interoperable applications in a variety of domains. The decentralized nature of compositions imply that they typically function in volatile environments where the parameters of the participating Web services change during execution. Thus, it is critical to adapt to these changes promptly and properly, so as to maintain a consistent and optimal composition. One traditional approach for adapting WSCs dynamically is to identify and query for the vital changes in the data of the various component services, and then integrate the revised data into the composition model. Queries are often costly and time-consuming, however, and must be carefully managed. In this thesis, I present a method that selectively queries component Web services for their revised values by using the value of changed information (VOC) approach. VOC measures the value of the change that revised information may potentially introduce to the composition. This metric is used to determine whether or not a query should be issued to a component service for its revised value. In doing so, we may focus only on those queries that obtain information that will greatly impact the composition and eliminate costly queries that are not needed. As computing VOC comes at a computational price, I present techniques for alleviating its complexity so that it becomes a more feasible approach. Furthermore, I present ways of generalizing the VOC-based querying approach so that it may be used to adapt a wide array of composition configurations as well as accomodate compositions that are sensitive to risk. In all, I demonstrate empirically and theoretically that the VOC-driven selective querying method compares favorably to other state-of-the-art querying approaches for adaptation.