ReSUS project (University of Stuttgart): Platform for reproducible research software basing on Dataverse

[doigl]

In the ReSUS-project we plan to build a platform to model, license, describe, pack together and automatically provision research software. Software will be packed together with all dependencies (e.g. in form of containers) and data in form of an ROAR-archive (enhancing the CSAR-format of the TOSCA standard
)). Dataverse will function as the repository to deposit the ROAR packages, describe them with appropriate metadata, link to a publication. Outside of dataverse will be components to model the dependencies (Winary), to recommend a suitable license dependent on the licenses of the dependencies (LicenseChecker) and automatic provision (OpenTosca-Runtime as an external tool).

Within Dataverse we could contribute to:

• metadata block for software/container
• git integration
• best practices on depositing code on Dataverse

and in addition:

• automated extraction of metadata (from git, from ROAR-archive)

We would be interested in:

• specific handling of software datasets in schema.org annotations (SoftwareApplication or SoftwareSourceCode instead of Dataset), DataCite-Metadata (type Software instead of Dataset) and Citation-Data
• synchronizing version of software and dataset

We have already implemented a software metadata block basing on CodeMeta in our Dataverse instance (https://darus.uni-stuttgart.de). Feedback from our researchers is quite good, the block is used (see for example https://darus.uni-stuttgart.de/dataset.xhtml?persistentId=doi:10.18419/darus-475). But we would like to enhance the metadata block with information about the "research" part of research software (what kind of research does this software enable, what model is implemented,...) and also with information about the runtime parameters of containers.

As use cases we have:

• Software as a primary research result: publish releases of medium research software projects in a citable and findable way (e.g. DuMuX, Espresso).
• Software as a means of research: publish data and code together in form of research object to enable reproducibility of a specific research result.

The project started in summer 2020 and will go for three years (project website).

Two other project handling research software can also provide use cases, input to metadata for software and integration with other external tools for software (Jupyter-Hub, VipLab). One of them is within the consortium for a national research infrastructure for engineering NFDI4Ing with a duration of 5/10 years, the other a local project started in 2019 for 3 years.

[4tikhonov]

Hi @doigl, we're very interested in CodeMeta schema as well, can you please share your GitHub repository with .tsv file? Do you have a plan to merge it in the Dataverse master branch?

[doigl]

Hi @4tikhonov,
sure, you can find the tsv here: https://github.com/TIK-NFL/dataverse/releases
I put some more information about the metadata block into the Google docs folder of the metadata working group. And I would very much like to merge it in the Dataverse master branch, but I thought, there will be some more discussion about and work on it in the working group.