RDataTracker and DDG Explorer - Capture, Visualization and Querying of Provenance from R Scripts

Scientific data provenance is gaining interest among both scientists and computer scientists. The current state of the art of provenance capture requires scientists to adopt new technologies, most commonly workflow systems such as Kepler [BML06], Vistrails [SKS08] or Taverna [MBZ08], among others. While there are likely additional benefits to adopting these systems, they present a hurdle to scientists who are more interested in focusing on science than in learning new technologies. The work described in this poster is aimed at exploring the extent to which we can support scientists while expecting a minimal investment in terms of additional effort on their part. This work has been developed in collaboration with ecologists at Harvard Forest, a 3500 acre facility operated by Harvard University and serving as a Long-Term Ecological Research (LTER) site funded by the National Science Foundation. Many of these ecologists perform data analysis using R, a widely used scripting language that includes extensive statistical analysis and plotting functionality. These scientists are committed to understanding their data, making sure that their data analyses are done in an appropriate manner, and sharing their data and results with others. For these reasons, they appreciate the value that collecting data provenance may have, but they are not enthusiastic about learning new tools. In this poster, we present two tools aimed at this audience: RDataTracker and DDG Explorer. RDataTracker [LB14] is used to collect data provenance during the execution of an R script. DDG Explorer is the tool that is used to examine and query the resulting data provenance.