Data Flow Modeling In Service Workflow

Service workflow was proposed as a means of combining a variety ofservices to provide more complex functions, and the formation of theBPEL specification, marks the study of service workflow has reached arelatively mature stage. However, the emergence of new applications givesnew challenges to service workflow, especially with the dataflowprocessing, such as the huge amount of data interaction in cloud computingand scientific computing and the requirement of new data format such asfiles; the integration of human activity may involve several output files;pervasive computing asks service process to be context-aware.In response to the above problems, this article studied the dataflowmodeling in service workflow, mainly include two aspects, the design ofdataflow model, as well as the establishment of the dataflow.In the aspect of dataflow model design, this article proposed acontext-aware dataflow model, to pass different data to activities indifferent context. The article first analyzes the definition and theperception of the context which activate dataflow, and then defines severaldata models used in different application scenarios, and finally, defines thedata mapping between input and output parameters of activities, enableworkflow engine to pass corresponding data to activities according to thedataflow status, whether it is activated or not.In the aspect of dataflow establishment, to solve the problem ofcomplex and error-prone data mapping between complex messages, this article propose an automatic data mapping method in service dataflow.Make use of the existing automatic schema matching techniques to matchthe service parameters, and combine the control flow information toimprove matching accuracy, and finally establish dataflow automatically.Two examples are illustrated to prove the proposed method.