The basis for versioning is an effort to meet the research requirements for reproducibility in our infrastructure.
Språkbanken Text's datasets and analyses ("resources") are either development versions or stable versions.
A development version is characterized by:
- we only publish the latest version
- we give the resource an identifier without a version number (e.g., superlim)
- we do not publish previous development versions
- our citations are timestamped with the latest modification date (compared to web links).
Currently, almost all of Språkbanken Text's resources are development versions.
A stable version is characterized by:
- we have decided that a version should be stable and assign a semantic version number to the identifier (e.g., superlim-2.0)
- we publish all stable versions
- our citations point to a specific stable version.
Our long-term ambition is to have at least one stable version for all of our datasets and analyses, to enable reproducibility of research supported by Språkbanken Text.
Stable versions are kept for at least 10 years.
Example of versioning
- sb-1 (stable version)
created: 2022-01-01 (date it was published)
updated: 2023-01-01 (when the development froze and became stable sb-1)
successor: sb-2 - sb-2 (stable version)
created: 2023-05-10 (date it was published)
updated: 2024-01-01 (when the development froze and became stable sb-2) - sb (development)
created: 2024-06-10 (date it was first published)
updated: 2025-01-25 (latest update of this version)