An Enabling Environment For E-Science-Oriented Exploratory Service Composition

Today, Service-Oriented Computing is becoming an important paradigm in distributedcomputing area. Service composition becomes the main way to construct service-oriented,loosely integrated application systems. Nowadays most service composition technologies areonly suitable for solving the problems that requirements are definite and business processesneed to be predefined. In many application areas such as scientific collaboration,tele-medical-services, city emergency and so on, people have realized it is hard to know thewhole process of how to compose the service-oriented resources to achieve their goals inadvance.Motivated by the requirements of the project for developing a dynamic problem-solvingenvironment for scientists, by which scientists can not predefine the whole application processbecause of the creative and uncertain nature of the scientific research, a new approach toexploratory service composition is brought forward. This approach emphasizes applicationconstruction (programming) at runtime. Using this approach, users could compose thoseend-user understandable and executable business services to develop applications on the fly.We developed an enabling software environment for the exploratory service compositionapproach. The software allows scientists to write workflows without having to learn theexploratory language. It has a number of important features: It supports application constructionin the course of execution;It provides a graphical user interface, all the resources and basicoperations are presented graphically;It provides semi-automatic application development. Usersonly need, based on predefined modeling elements and rules, to drag and configure the symbolsrepresenting services to express their specific requirements;It provides a real-time monitor.Through the monitor, users could watch the process of services' execution and results of them.This facilitates users to obtain exceptions in time and deal with them in time;It provides ajust-in-time service recommendation mechanism. This mechanism is targeted at assistingscientists by recommending the most promising services that will be required.Using the above development system and the real requirements of the "Network ScientificResearch Collaboration" project as a test bed, we proved the feasibility and validity ofexploratory service composition approach.