Glossary Term Protocol (GTP) Specification

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

Overview

The Glossary Term Protocol (GTP) defines the structure, exposure, and trust scaffolding of glossary terms published as machine-ingestible fragments.

GTP is used to anchor defined terms referenced throughout a Semantic Digest, enabling AI agents and retrieval systems to:

Resolve term meanings unambiguously
Cite source-aligned definitions
Retrieve term fragments in multiple formats (TTL, JSON-LD, PROV, etc.)
Map terms to external ontologies (e.g., WikiData QIDs)

Each GTP-compliant term is published as a DefinedTermFragment, embedded in HTML or exposed via digest endpoints.

Fragment Format

Terms are embedded in HTML using SDT with data-sdt-class="DefinedTermFragment", like this:

<template data-visibility-fragment data-sdt-class="DefinedTermFragment">
---
data_id: term-mooptotal
defined_term: Maximum Out-of-Pocket
definition: The total amount you could pay in a year for covered services.
source: https://medicare.gov/moop-definition
glossary_scope: medicare-advantage
graph_alignment:
  sameAs: https://www.wikidata.org/wiki/Q123456
  property: P1234
export:
  - jsonld
  - ttl
  - prov
...
</template>

Required Fields

Field	Description
`data_id`	Unique identifier for the term fragment
`defined_term`	Canonical label (e.g., “Part B Premium”)
`definition`	Human-readable definition text
`source`	Canonical source or reference URL

Optional Fields

Field	Purpose
`glossary_scope`	Domain or topic scope (e.g., `medicare-advantage`)
`graph_alignment.sameAs`	Link to WikiData, DBpedia, or ontology URI
`graph_alignment.property`	Optional vocabulary predicate (e.g., `P6789`)
`export`	List of supported formats
`prov`	Whether W3C PROV metadata is enabled

Digest Endpoint Access

Each GTP term is also accessible as a content-negotiable Semantic Digest:

GET /semantic/json/term-mooptotal
GET /semantic/ttl/term-b-premium

Or using query string negotiation:

GET /semantic/term-b-premium?format=ttl

Each endpoint exposes:

@type: DefinedTerm
@id matching the data_id
Full provenance metadata (if prov: true)

Usage in SDT Fragments

Any SDT (e.g., DataFragment, FAQFragment, DerivedStatsFragment) may reference a defined term via the glossary_id field:

glossary_id: term-mooptotal

This enables:

Term resolution
Definition linking
AI/agent-level consistency in retrieval

External Alignment (sameAs)

GTP supports sameAs mappings for knowledge graph and LLM alignment:

graph_alignment:
  sameAs: https://www.wikidata.org/wiki/Q7654321

These help reinforce entity understanding across:

LLM memory
Knowledge graphs
Co-citation behavior

Trust and Provenance

Each GTP term may optionally include PROV metadata:

prov: true
source: https://data.cms.gov/...

This allows each definition to:

Declare its origin
Be independently verifiable
Align with provenance claims in the digest

Registry Behavior

GTP-compliant terms may be:

Embedded inline (via SDT)
Published as full digests (e.g., /semantic/json/term-mooptotal)
Federated via .well-known manifest or GitHub registry