ProvenanceMeta Specification

Part of the WebMEM Protocol
Last Updated: 2025-07-28

Overview

ProvenanceMeta is the trust layer of the WebMEM Protocol. It enables each fragment or digest to expose machine-readable metadata that proves:

Where the data came from
Who published it
When it was created
How it was derived or modified

SDP uses the W3C PROV standard as its foundation, with additional custom fields for domain-specific citation and source traceability.

Purpose

Provenance metadata enables AI agents, validators, and retrieval systems to:

Trust the source of a fragment or term
Cite the originator accurately
Validate the content’s authority
Differentiate primary vs secondary data

This is critical in high-trust domains such as healthcare, legal, scientific, or regulatory publishing.

Activation

You can activate provenance in any SDT or GTP fragment via:

prov: true

This triggers inclusion of a provenance block during digest rendering.

Core Fields

Field	W3C PROV Mapping	Description
`prov:wasDerivedFrom`	✅	Canonical source URL or dataset identifier
`prov:wasAttributedTo`	✅	Person, org, or system that authored/published the content
`prov:generatedAtTime`	✅	Timestamp of content or digest creation
`prov:alternateOf`	✅	Alternate representation or upstream content ID

Example

prov: true
source: https://data.cms.gov/moop-2025
provenance:
  prov:wasDerivedFrom: https://data.cms.gov/downloads/MOOP.csv
  prov:wasAttributedTo: https://medicarewire.com
  prov:generatedAtTime: 2025-06-29T14:08:00Z
  prov:alternateOf: https://cms.gov/moop-guidance

Export Behavior

Provenance fields are automatically included in digest formats that support it:

✅ Turtle (TTL)
✅ JSON-LD (@context includes prov:)
✅ W3C PROV XML (application/provenance+xml)

They are omitted from Markdown and CSV unless explicitly rendered via export flags.

Attribute Shortcuts (HTML Binding)

When fragments are embedded using HTML data-* attributes, key provenance fields can be flagged:

<span
 data-digest="/semantic/json/term-mooptotal"
 data-source="CMS.gov"
 data-prov="true"
 data-entity-id="term-mooptotal">
 Maximum Out-of-Pocket
</span>

This allows crawlers and AI agents to:

Resolve trust context directly from page elements
Retrieve structured provenance from the digest

Best Practices

Always include prov:wasDerivedFrom with a canonical source
Use prov:wasAttributedTo to declare authorship or publishing organization
Timestamp content with prov:generatedAtTime if stable
Use alternateOf to point to an upstream reference or document version