User-Submitted Annotations on Provenance

Printer-friendly version

Motivation

Provenance can be used to answer user questions about the steps taken to generate a data product. Many eScience-based systems will generate data products for later review by end users. For these end users, establishing an understanding of how data products were created is useful for varying reasons. This makes provenance tracking for data products an important component for eScience systems.

Despite playing this important role, the information in a provenance record alone may be insufficient to ensure an end user’s understanding of how the corresponding data product came to be. Additional information, which falls outside of the traditional scope of provenance, can and often will be necessary to accomplish this. Examples of such information include:

  1. Definitions of provenance steps, consumable by end users
  2. Contextual information that ends up impacting final results. As an example, for data gathering at an observatory (as with our SPCDIS case study), cloud cover may impact data gathering, but not be explicitly recorded in provenance.

Given that supplemental information can be useful for provenance records, how should a design team for an eScience system decide what to provide? This can be a challenging problem, since it may not be clear which kinds of additional information would benefit end users. Certain information may end up never being used by end users, while other kinds may be needed that didn’t occur to the design team. Without insight into what kinds of supplemental information are needed by end users, a development team may waste significant time generating information irrelevant to the needs of an end user community.

Objective of Work

At present, the goals of this work are as follows:

  • Identifying specific ways in which provenance could be extended by user-submitted annotations, driven by eScience-oriented use cases
  • Development of a new module for the Proof Markup Language (PML), designed to capture user annotations on provenance
  • Design and development of interfaces for enabling users to create and view annotations on provenance records.

See Also