How to contribute to UD
Current Data
- Obtaining data: The latest official version of UD treebanks can be downloaded from LINDAT/CLARIN. If you prefer to access treebanks via GitHub repositories, make sure you work with the master branch, which corresponds to the latest official release.
Documentation and Discussion
-
Feedback and discussion: The issue tracker provides a forum for discussion of UD guidelines or treebanks. Any issues related to documentation and guidelines, even specific to one language, belong to the issue tracker of the docs repository. Issue trackers of individual treebank repositories should be used only to report bugs in those treebanks. You can also send an e-mail to the UD mailing list; however, the list is primarily meant for announcements for data maintainers.
-
Language-specific guidelines: Language-specific documentation pages can be modified via the edit link at the top of the page. There are also technical instructions for how to create language-specific documentation.
-
Changes to general guidelines: Guidelines change history
Contributing Data
- New contributions: If you want to start a treebank or contribute to a release, please see:
- Advice on how to start a new treebank
- Technical steps in the release checklist
-
Data corrections: If you do not want to commit to maintaining a treebank but you want to help fix a particular bug in the data, you can submit a pull request against the dev branch of its GitHub repository. The master branch should not be modified as it is built automatically by the release pipeline.
- Data validation: Every treebank must comply with general and language-specific validation rules in order to be included in the official release. The on-line validation report provides a dashboard with treebank statuses. The tools repository can be cloned and for running validation locally (note that data in this repository is automatically updated based on the language-specific lists.
Email list
Please consider subscribing to the UD mailing list, especially if you are going to contribute data. Important announcements for the data providers are circulated through this list.