Behavioral Analysis of Scientific Workflows With Semantic Information

Abstract

The recent development in scientific computing related areas has shown an increasing interest in scientific workflows because of their abilities to solve complex challenges. Problems and challenges that were too heavy or time consuming can be solved now in a more efficient manner. Scientific workflows have been progressively improved by means of the introduction of new paradigms and technologies, being the semantic area one of the most promising ones. This paper focuses on the addition of semantic Web techniques to the scientific workflow area, which facilitates the integration of network-based solutions. On the other hand, a model checking technique to study the workflow behavior prior to its execution is also described. Using the unary resource description framework annotated Petri net formalism, scientific workflows can be improved by adding semantic annotations related to the task descriptions and workflow evolution. This technique can be applied using a complete environment for the model checking of this kind of workflows that is also shown in this paper. Finally, the proposed methodology is exemplified by its application to a couple of known scientific workflows: the first provenance challenge and the InterScan protein analysis workflow.