You are here

Harvard Forest >

Harvard Forest REU Symposium Abstract 2019

  • Title: provBookR: Visualizing Data Stories
  • Author: Erick Oduniyi (University of Kansas)
  • Abstract:

    How scientific results come to be is the product of various domain-specific, economic, historical, social, and technical factors. As a consequence, information about how these factors shaped and produced these results is essential to both artistic trade and scientific reproducibility. Whether these results are material or digital, provenance provides the historical accounts of objects: paintings, bones, essays, scientific tables, and plots. Using the idea of provenance and the associated tools that collect provenance, we develop an R package that presents visualizations of provenance for R scripts. This tool is called provBookR, and it allows users to get the history of single R data objects (e.g., plots, scientific tables, variables). provBookR accomplishes this by utilizing previously developed R provenance tools. In particular, provBookR relies on RDataTracker (rdtLite) to record provenance generated from scripts in R, provGraphR for querying the collected provenance, and provParseR for processing the provenance. As a result, provBookR is capable of recording and generating presentations of data provenance from individual R data objects, which users specify through a browser interface. The presentation done by provBookR is in the form of digital booklets called provbooks — interactive web pages created from R. The motivation for provBookR's booklet interface is to produce a document with familiar appearance containing the data provenance and animations of the data provenance. The animations and visualizations represent the various data manipulations of the respective data objects. In doing so, provBookR affords scientists and programmers a visual aid for understanding and reproducing R scientific analyses and results.

  • Research Category: Group Projects