Uma abordagem baseada em padrões para o intercâmbio entre especificações de workflows científicos

Abstract

Scientific workflows have been used to solve complex problems in different areas. Scientific Workflow Management Systems (SWfMSs) are used for specifying and managing these workflows. Nevertheless, each SWfMS may have different characteristics and its own workflow specification language, making its reuse accross different SWfMSs a difficult process. The lack of semantic standardization makes this reuse even more difficult, since the workflow modeling elements available in some SWfMSs may not be mapped onto others SWfMSs. The use of an intermediate language for the interchange of scientific workflows may help with the reuse of workflows developed in different SWfMSs, as it allows for the definition of a common framework for these SWfMSs. Nonetheless, such a language does not prevent the loss of some semantic information during a specification transformation process between different SWfMSs, since this language must be robust enough to represent the semantics of diverse workflows developed in different SWfMSs. The identification of scientific workflow patterns may help to describe the most important semantic information for the construction of these workflows. Thus, the purpose of this study is to provide a pattern-based approach for the interchange of scientific workflow specifications, using an intermediate language that supports semantic information obtained through the description of workflow patterns. This thesis also analyses the results obtained with the proposed approach being applied to workflow specifications stored in the myExperiment repository.