Review algebra is implementation-independent
The same review object can be edited by
dplyr
Excel
Shiny
OpenRefine
Review rounds as provenance activities
Show
candidate ↓ review_1 ↓ review_2
and explain that every review round corresponds to one provenance activity.
Provenance attributes
For example
attr(gdp, "prov_activity") candidate review_1 review_2 download impute forecast
attr(gdp, "prov_agent")
attr(gdp, "prov_used")
Relationship to PROV
Without implementing RDF.
Simply explain
| Review algebra | PROV |
|---|---|
| candidate/review columns | Entity |
| prov_activity | Activity |
| prov_agent | Agent |
| prov_used | used |
| review chain | wasGeneratedBy |
The important point is
The package stores provenance in a lightweight R representation that can later be serialised into PROV or other provenance models.