Research software documentation

Guideline 12(interner Link) defines the comprehensive documentation of information relevant to the development of research findings as being a fundamental element of good research practice. Ensuring verifiability, reproducibility and replicability are key aims in connection with the use and development of research software, too. For this reason, every piece of research software used should be fully included in the documentation of project results (including version information, parameter settings, test/training datasets, etc.) and cited in accordance with subject-specific conventions. This documentation should be embedded in the research process and should be verifiable. It should be published together with or as part of the written documentation so that these results can likewise be assessed by the research community.

There may be cases in which research software cannot be fully documented for technical or disciplinary reasons, such as in cases where commercial and in-house components are combined, or where there is a potentially unlimited number of descriptive parameters. Here, it is crucial to document the properties and parameters of the software relevant to the research question to be documented – also including the source code itself wherever possible. In order to ensure effective documentation and citation of self-developed research software, metadata should be generated in a way that is as automated and standardised as possible. This includes linking it to electronic notebooks or lab books and selecting appropriate publication platforms for software. Once a certain technological level has been reached, or where there is significant relevance for users or a broader scope, it is advisable to register the research software.

For example in the Research Software Directory(externer Link) or the Helmholtz Research Software Directory(externer Link).