Master Thesis
Title: Automated Service Synthesis in a Weighted Framework.
Abstract:
Service Oriented Computing (SOC) is a computing paradigm that uses services as its building blocks, where services are software modules which are self-described, self-contained, platform independent, and available via networks. SOC promises a rapid, low-cost, interoperable and evolvable application development. One important concept in SOC is software re-usability. When the requested service is not available, (part of) available services can be composed into a new service (called target service) in order to fulfill the request. The process of composing the available services is called service composition. One important process in service composition is service synthesis. Service synthesis constructs the specification that specifies the composition of available services in order to obtain the target service. This process can be done either manually or automatically. Numerous researches have been conducted to automate such a process, but complete automation has not been achieved yet. The Roman Model is one of the proposed frameworks for automatic service synthesis. The key feature of the Roman Model is its capability to capture and abstract the service behavior in terms of possible evolutions of the service and supported interaction of the service with the client (human or another service). It represents the service using transition System. Moreover, it provides a sound and complete algorithm for synthesizing all possible compositions of available services based on simulation checking. In many situation, not all compositions can be considered as equally good, e.g., the execution of one service might be faster or require less resources (bandwidth, memory, etc.) than that of another one. In these cases, one might be interested in understanding whether there is a "best" composition and if yes, in synthesizing it. In this thesis we extend the Roman Model such that it is able to model the best service composition synthesis problem obtaining the Weighted Roman Model. We use Weighted Transition Systems (WTS) to model the service, so that we can capture the cost of operation execution done by a service. Hence within this setting, we can make a comparison among all compositions of the available services by considering the total cost of operation execution performed by each possible composition. Furthermore, we also propose algorithm for synthesizing the best composition among all possible compositions and show that it is sound and complete.