Datasets
ActiveA dataset record describes data; it does not contain it. The data itself lives wherever it is large enough to need to live, and the record says where, in what version, and under what licence.
| ID | Dataset | Source | Size | Licence | Used by |
|---|---|---|---|---|---|
| DATA-001 | Placeholder | Not set | Not set | Not set | EXP-001 |
| DATA-002 | Placeholder | Not set | Not set | Not set | EXP-002 |
Why data is not in this repository
This repository is documentation. Committing a multi-gigabyte capture to it makes every clone slow forever and still does not make the data citable. The record here carries the identifier, the version, the checksum and the location; that is what makes a result reproducible, not proximity.
Binary formats commonly used for research data are refused by .gitignore for
this reason.
What every record must contain
| Field | Why |
|---|---|
| Provenance | Where it came from and who may use it |
| Schema | What each field means and its units |
| Collection method | How it was produced, including what was excluded |
| Preprocessing | Every transformation between raw and used |
| Quality | Known defects, missingness, and their effect on results |
| Licensing | What may be redistributed, and what may not |
| Location | Where the bytes are, with a checksum and a version |
Licensing check
Before any dataset is used in a publication, confirm that the licence permits the use and the redistribution the venue requires. Discovering at camera-ready that data cannot be released is a costly and entirely avoidable failure.