On User Views in Scientific Workflow Systems

An increasing number of scientific workflow systems are providing support for the automated tracking and storage of provenance information. However, the amount of provenance information recorded can become very large, even for a single execution of a workflow - (6) estimates a ten-fold blowup of the size of the original input data. There is therefore a need to provide ways of allowing users to focus their attention on meaningful provenance information in provenance queries. We highlight recent work in this area on user views, showing how they can be efficiently computed given user input on relevance, or and how pre-existing views can be corrected to provide accurate provenance information. We also discuss how to search a repository of workflow specifications and their views, returning workflows at an appropriate level of complexity with respect to a hierarchy of views.