The Dataverse project

Dataverse (Dataverse) is an open source web application to share, preserve, cite, explore and analyze research data. Researchers, data authors, publishers, data distributors, and affiliated institutions can receive appropriate credit via a data citation with a persistent identifier (e.g., DOIs). Many universities and research institutions have Dataverse installations. For example, the Harvard Dataverse, open to all scientific data from all disciplines worldwide, or DataverseNL, located in the Netherlands providing data management services for a number of Dutch Universities.

A Dataverse site hosts multiple dataverses. Each dataverse contains datasets or other dataverses, and each dataset contains descriptive metadata and data files (including possibly documentation and code that accompany the data). The Glossary delineates core Dataverse terminology from core DataLad concepts.

Goal of this DataLad extension package

This DataLad extension package provides interoperability with Dataverse for the purpose of depositing DataLad dataset on Dataverse, and for retrieving DataLad datasets from Dataverse instances, together with their full version history. It comes with several features that enable the following main use cases:

  1. Deposit file content for any number of file versions tracked in a DataLad dataset on dataverse, including the dataset's history, for retrieval with DataLad

  2. Export a single version of a DataLad dataset to Dataverse, with all files organized in a matching human readable directory tree, for consumption via the Dataverse web UI, or tools other than DataLad

Typically, DataLad workflows for publishing data to external providers involve some repository hosting service (like GitHub or GitLab). This service holds the "Git" or version-control-system part of the dataset: dataset history (commit messages), non-annexed files (usually code, text), and file identity information for all annexed files (file name, but not content). These workflows will often pair the repository hosting service with a non-specialized hosting service (like Dropbox or AWS S3), which doesn't understand Git, but can hold annexed file content (however, some services are able to store both parts, e.g. GIN):

DataLad publishing: git and git-annex parts are separate

With datalad-dataverse, the entire dataset can be deposited at a Dataverse site, forming a so-called dataset sibling. Internally, this is achieved by packaging the "Git" part and depositing it alongside the annexed data, similar to how the datalad-next extensions allows to do this for webdav based services.

The primary use case for Dataverse dataset siblings is dataset deposition for example for the purpose of data preservation, and the possibility to cite a dataset via Dataverse's persistent identifiers. Typically, only one site will upload a dataset and file content (updates) for others to reuse. Compared to workflows which use repository hosting services, this solution will be less flexible for collaboration (because it's not able to utilize features for controlling dataset history offered by repository hosting services, such as pull requests and conflict resolution), and might be slower (when it comes to file transfer). What it offers, however, is the ability to make the published dataset browsable like regular directories and amendable with metadata on the Dataverse instance while still being cloneable via DataLad.

What can I use this extension for?

You can use this extension to publish and share your dataset via Dataverse, and you can use it to clone published DataLad datasets from Dataverse. Here is some inspiration on what you could do:

  • Publish your study (including its version history, data, code, results, and provenance) as a DataLad dataset to Dataverse to share it with collaborators

  • DOIify your work by getting a DOI for it from Dataverse.

  • Share a published dataset's URL with colleagues and collaborators to give them easy access to your work with a single datalad clone.

  • Clone a friend's DataLad dataset -- from Dataverse!

The combination of DataLad and Dataverse enables the additional use case of a data-less dataset deposition. A DataLad dataset can be conceptualized as an actionable metadata collection with precise information on identity and availability of all components of a dataset. So precise, that DataLad can use this information to retrieve any component from any physical storage location, and verify its identity. So even without depositing a physical copy of all data associated with a dataset, a DataLad dataset on Dataverse is a precise and verifiable deposit that allows for the location of its components to change, but not their content or identity.

What can I not use this extension for?

  • This particular extension package does not work for datasets stored in a service other than Dataverse. Please refer, for example, to the list of special remotes as hosted by the git-annex website for other storage services and how to use them with DataLad, and to the DataLad handbook for an overview of other extension packages.

  • Dataverse installations may have upload or storage limits. Exceeding those limits is not possible with this tool. However, you will be able to at least deposit the revision history of your dataset even if dataset content is too large.

  • The starting point for working with this extension is a DataLad dataset, not only a collection of files in a directory. This extension package will also not transform existing Dataverse dataset into DataLad datasets, but expose DataLad datasets as Dataverse datasets.

  • Please see the feature support section for details on what is and is not supported by this extension package.