Adding Ontologies to Scientific Workflow Composition

Scientific workflows are being used as an abstraction for the composition of large scale scientific experiments. As scientific workflows become more complex, these abstractions isolate scientists from infrastructure issues. Although representing a workflow in an abstract level is a first step, there are many open issues, such as the ones related to semantics. Adding semantics to abstract workflows enables the explicit representation of which activities can be linked to each other, or which activities are equivalent to each other. However, representing an abstract workflow with semantics is an open problem. Existing approaches address either the representation of abstract workflows or the use of domain ontologies to add semantics to activities, but not both. In the latter case, these approaches focus only on adding semantics to executable workflows, instead of working in different abstract levels. This makes it difficult to group workflows into a common abstract representation in the conceptual level. Common abstract workflow representation allows for a better understating of the experiment by scientists and enables the association of the experiment definition with domain knowledge. This paper proposes coupling workflow ontologies to abstract workflow representations. This provides the required semantic mechanisms to help scientists to identify equivalent activities, group executable activities into one abstract activity with the same semantics, and verify the compatibility between activities. We implemented and evaluated the proposed approach by coupling SciFlow – a workflow ontology with GExpLine – an abstract workflow management tool. Experiments show the benefits of this approach.activity with the same semantics, and verify the compatibility between activities. We implemented and evaluated the proposed approach by coupling SciFlow – a workflow ontology with GExpLine – an abstract workflow management tool. Experiments show the benefits of this approach.