How can a dataset expose a version and persistent identifier?

Definition: A dataset can expose version for the version embodied by the described resource and identifier for an identifier assigned to the dataset. In Schema.org, version accepts a number or text, while identifier accepts a PropertyValue, text, or URL.

Key facts

• Google says datasets are easier to find in Dataset Search when pages provide supporting information such as name, description, creator, and distribution formats as structured data, according to its Dataset structured data documentation.
• Schema.org defines version as the version of a CreativeWork embodied by a specified resource and lists Number and Text as its value types in the version property definition.
• Schema.org says that identifier can represent any kind of identifier for any Thing, including ISBNs, GTIN codes, and UUIDs, in its identifier property definition.
• The Schema.org Dataset definition shows version and identifier in a JSON-LD dataset example.

What does version describe?

Google's Dataset documentation points to Schema.org's Dataset definition for the full definition of Dataset. The version property describes the version of a CreativeWork embodied by a specified resource. Its value types are Number and Text, so a dataset can publish a numeric release label or a textual label.

The Schema.org Dataset example uses a text value, 2013-11-21, for version. That value is part of the dataset description and is separate from the identifier object shown below.

{
  "@type": "Dataset",
  "version": "2013-11-21",
  "identifier": {
    "@type": "PropertyValue",
    "propertyID": "https://registry.identifiers.org/registry/doi",
    "value": "doi:10.5066/F7VX0DMQ",
    "url": "https://doi.org/10.5066/F7VX0DMQ"
  }
}

Which identifier values can describe dataset identity?

Schema.org's identifier property definition lists PropertyValue, Text, and URL as its value types. It describes the property as an identifier for a Thing and says that identifiers can be represented as textual strings or URL links. A PropertyValue object is useful when the markup needs to carry a scheme in propertyID alongside the value.

Google's source and provenance guidance says to use identifier for relevant Digital Object Identifiers, or DOIs, and Compact Identifiers. When a dataset has more than one identifier, the documentation says to repeat the identifier property. In JSON-LD, it represents that repetition with JSON list syntax. The Dataset example uses a PropertyValue for a DOI and includes a DOI URL.

How does a version relate to copied or republished datasets?

Google's source and provenance guidance recommends structured data on the canonical page when a repository has a landing page and pages that list several datasets. For a simple republication, its guidance uses sameAs for the most canonical URL and requires the value to identify one dataset unambiguously. The same value should not be used for two different datasets. Google recommends isBasedOn when the republished dataset or its metadata has changed significantly, and also when a dataset derives from or aggregates several originals.

This gives the fields separate jobs: identifier carries the dataset identifier, version describes the version embodied by the resource, and sameAs links a simple republication to a canonical URL. The three properties can appear on the same Dataset node when the page describes both a release and its provenance.

How can you check the markup?

Google's Dataset documentation recommends validating the JSON-LD with the Rich Results Test, then using the URL Inspection tool to test how Google sees a page. The page must be accessible to Google and cannot be blocked by a robots.txt file, a noindex tag, or login requirements. Check that the version value belongs to the described resource, that each identifier value is the intended DOI, Compact Identifier, text, URL, or PropertyValue, and that a repeated identifier is represented as a JSON-LD list.

Related terms

• Dataset structured data
• Structured data
• JSON-LD
• Schema.org @id

FAQ

Can you use a date as a dataset version?

Yes. Schema.org lists Number and Text as the value types for version, and its Dataset example uses the text value 2013-11-21.

Can you publish more than one dataset identifier?

Yes. Google's documentation says to repeat identifier when a dataset has more than one identifier and to use JSON list syntax in JSON-LD. A value can be a DOI or Compact Identifier, and Schema.org also permits Text, URL, or PropertyValue values.

Sources

1. Google Search Central, "Dataset structured data", 8 September 2026
2. Schema.org, "Dataset", n.d.
3. Schema.org, "version", August 2026
4. Schema.org, "identifier", August 2026

---

Check how a page reads to answer engines. Open the GEO score analyzer