Applying Workflows to Scientific Projects Represented in File System Directory Tree
TimeWednesday, 11 November 20201:30pm - 2pm EST
DescriptionMost workflow management systems represent scientific workflows as DAG (directed acyclic graph) and require to provide a workflow specification before the workflow can be executed. Workflow can be designed using a graphical interface or with textual workflow editing. For uses unacquainted with workflow abstraction, these systems can be complicated as it usually takes multiple steps to design abstract workflow specification. To reduce the learning curve and barriers to entry for inexperienced users, we have decided to create a workflow representation that is based on the well-known file system directory structure. In our representation workflow structure can be drawn from project organization space which is based on the file system directory tree. As a result, proposed workflow representation is more natural and more intuitive for users who are already accustomed to the directory structure. The workflow structure is created on the fly as the user adds, executes and organizes services inside their workspace. We have used tree structure hierarchy to model a part-whole hierarchy between workflows, sub-workflows and individual services. Proposed solutions support modularity with multi-level sub-workflow nesting, encapsulation, workflow reuse and sharing, as well as automated workflow execution in a distributed infrastructure. The proposed model was implemented and evaluated on the IS-EPOS Platform, a real-world online workflow management system used to conduct research in the field of anthropogenic seismicity.