Semantic Software Metadata for Workflow Exploration and Evolution

Abstract

Scientific workflow management systems play a major role in the design, execution and documentation of computational experiments. However, they have limited support for managing workflow evolution and exploration because they lack rich metadata for the software that implements workflow components. Such metadata could be used to support scientists in exploring local adjustments to a workflow, replacing components with similar software, or upgrading components upon release of newer software versions. To address this challenge, we propose OntoSoft-VFF (Ontology for Software Version, Function and Functionality), a software metadata repository designed to capture information about software and workflow components that is important for managing workflow exploration and evolution. Our approach uses a novel ontology to describe the functionality and evolution through time of any software used to create workflow components. OntoSoft-VFF is implemented as an online catalog that stores semantic metadata for software to enable workflow exploration through understanding of software functionality and evolution. The catalog also supports comparison and semantic search of software metadata. We showcase OntoSoft-VFF using machine learning workflow examples. We validate our approach by testing that a workflow system could compare differences in software metadata, explain software updates and describe the general functionality of workflow steps.