Contributions or parts of contributions from scientific investigations are often released as separate semantically annotated datasets for several reasons, for instance, to enable the development of machine learning applications to automatically identify contributions for unseen articles. A concrete example of such a dataset for import in the ORKG would be SoMeSci -- a dataset of scholarly articles annotated with their software mentions either in terms of usage toward the contribution of the work or created as a contribution of the work itself. This help article delineates the steps users can follow to import their related datasets on themes across Science at large in ORKG.

To import your dataset in ORKG, the following general steps are offered as guidelines.

1. Define your data model. Note the ORKG graph begins as follows Article -> has contribution -> Contribution1 -> your data. Thus the data model will encapsulate your data. Concrete examples of previously created data models are here and here.
2. Next, if your dataset specifies one-level contribution graphs datasets, we recommend using the CSV import feature to bulk import papers where each paper can be specified as rows in a CSV file. Please note: To avoid browser crashes during import, it is advisable to import your data in batches of approximately 100 rows per file per import.
3. Otherwise, for nested contribution graphs datasets, we recommend using the ORKG Python API. For this, basic python scripting knowledge is necessary. With the ORKG Python API, datasets of contributions in multiple levels of nested graphs can be easily imported into the Open Research Knowledge Graph. The following python script is an example which utilizes several components of the ORKG Python API to bulk import a dataset created as per this data model.

For further information or support in importing your datasets please contact us on info@orkg.org and a team member will be happy to help you get started!