Dataset, validation and exports
Exports and the audit trail
What the FoodProt AI export contains, who can export, and what the provenance columns let a reader check.
Approved values only
An export contains values a reviewer approved. Candidates, rejected values and values with open flags are not in the file, and no option adds them.
Open errors in the verified dataset block an export, and the export page says how many there are and where to clear them.
Who can export
Owners can export. An owner can also switch on "allow members to export" for the project, which lets editors and reviewers export too. Viewers cannot export. An export leaves the platform, so the owner decides who may produce one.
Two formats
The long-form dataset, one row per measurement, is the main scientific export. The legacy FoodProt workbook is an optional view of the same values and includes only measurements that fit one of its columns, so the long form is the one to cite.
Provenance columns
Every exported row carries its provenance: the article title, authors, year, journal and DOI; the database and query that found it; the screening decision and reason; where the value was read, with page, table or figure; the reported value and unit; the converted value, unit and conversion; and the reviewer and review date.
These columns let someone outside the project check any single number without asking you.
Exports do not change
Each export is saved once, together with the study configuration and the versions of the checks, units and vocabulary it was produced under. Later edits to the project do not change a file that has already been issued.
Exports are stored privately and downloaded through short-lived links; there is no public link to an export.
Related
The verified dataset
What a verified value is, how the dataset is organised, and how to filter it.
Data dictionary
Every field in the FoodProt AI export, listed in the order it is written.
Scientific validation
Measuring accuracy against a reference set built by hand, with example sets kept separate from real ones.
Project roles: owner, editor, reviewer, viewer
What an owner, an editor, a reviewer and a viewer can each do in a project, and how platform roles differ.