Deep dive into DOIs

[diegoripley]

I know that StatCan can create DOIs, so can Zenodo. How can I do it? I want to create datasets that can be referenced in journal articles. If process is too complicated, maybe make my own?

[diegoripley]

This is what Grok says:

DOIs are unique, persistent identifiers that make it easier to cite and link to digital objects like datasets, ensuring they're findable and citable over the long term. However, minting (creating) your own DOIs isn't something individuals can do directly—it's managed through official registration agencies. I'll explain how to get set up and some practical alternatives.

1. Understand the DOI System

• DOIs are issued by the International DOI Foundation (IDF) through authorized Registration Agencies (RAs). Common RAs include:
  • DataCite: Ideal for datasets, research data, and open repositories. This is probably the best fit for your project.
  • Crossref: More focused on scholarly publications, but they can handle datasets too.
• To mint DOIs, you (or your organization) need to become a member of an RA or work through an intermediary that is already a member. This involves fees, technical setup, and adherence to metadata standards.

2. Steps to Mint Your Own DOIs

If you want full control over creating and managing DOIs for your datasets (e.g., assigning them directly on dataforcanada.org), here's how to proceed:

• Choose a Registration Agency:

  • Go with DataCite if your focus is datasets. They're nonprofit, researcher-friendly, and support open data initiatives.
  • Visit datacite.org to learn more.

• Become a Member or Find a Sponsor:

  • Direct Membership: If dataforcanada.org is part of an institution (e.g., university, nonprofit, or government body), apply for membership. Fees start around $1,000–$2,000 annually (depending on your organization's size and location), plus per-DOI costs (often ~$1 per DOI).
  • Sponsored Membership: If you're an individual or small project, you can't join directly. Instead, partner with a DataCite member organization (e.g., a library, university, or consortium like the Canadian Research Knowledge Network). They can sponsor you and handle DOI minting on your behalf. Search DataCite's member list here for Canadian options.
  • Requirements: You'll need to provide structured metadata for each dataset (e.g., title, creator, description, license) in formats like DataCite Schema or Dublin Core.

• Set Up Technically:

  • Integrate with the RA's API (e.g., DataCite's MDS API) to automate DOI creation. This might require some development work on your site—tools like Python's datacite library or REST APIs can help.
  • Ensure your datasets are hosted persistently (DOIs must point to stable URLs that don't change).
  • Once set up, you can mint DOIs like 10.1234/yourdataset and add them to your site's metadata or landing pages.

• Timeline and Costs: Approval can take a few weeks to months. Budget for setup fees, annual membership, and any development costs.

If this sounds like overkill for a smaller project, there are easier alternatives below.

3. Easier Alternatives for Assigning DOIs to Datasets

If you don't need to mint DOIs directly on your site, use free or low-cost services that handle the heavy lifting. These integrate well with open data projects and will give your datasets DOIs without you needing membership:

• Zenodo (zenodo.org):

  • Free, open repository backed by CERN and integrated with DataCite.
  • Upload your datasets, and Zenodo automatically assigns a DOI. You can link back to dataforcanada.org in the metadata.
  • Great for versioning—each update gets a new DOI version.
  • Bonus: Integrates with GitHub for automated uploads if your data is in a repo.

• Figshare (figshare.com):

  • Free for individuals; assigns DOIs via DataCite.
  • Upload datasets, add metadata, and get a citable DOI. Supports embargoes and private sharing if needed.
  • They have institutional plans if your project grows.

• Dryad (datadryad.org):

  • Focused on research data; assigns DOIs for a small fee (~$120 per dataset, waived for some open access integrations).
  • Good for scientific datasets with curation support.

• Dataverse (dataverse.org):

  • If you want to host your own repository, set up a Dataverse instance (it's open-source). Many installations (e.g., through Scholars Portal in Canada) integrate with DataCite for DOI minting.

These services ensure your DOIs are registered properly and discoverable in systems like Google Scholar or DataCite's search.

4. Best Practices for DOIs on Your Datasets

• Metadata: Always include rich details (e.g., via schema.org or DataCite format) to make datasets searchable.
• Persistence: DOIs should point to a stable landing page with download links, not directly to files (in case URLs change).
• Licensing: Use open licenses like Creative Commons to encourage reuse.
• Testing: Start with one dataset on a free service like Zenodo to see how it works before scaling.
• Canadian Resources: Check out organizations like the Digital Research Alliance of Canada or Compute Canada—they might offer guidance or sponsorship for data projects.