Link: Fine-grained and efficient lineage querying of collection-based workflow provenance

This paper discusses how Taverna supports collection-based provenance. In Taverna, transformations are treated as black boxes. When a list of items of the transformation’s expected input data type is given as input to a transformation, Taverna processes each element of the list separately through the transformation.

The naive approach to collection-based provenance is to store each of these transformation invocations separately and then to perform queries on this trace. The approach suggested by the paper is to store workflow provenance instead so that performance does not degrade as the size of the trace becomes large.