Reference

People Group Model

The Global.Church knowledge graph models people groups through four layered classes. Each carries a different question:

gc:PeopleGroup — the classification concept. Joshua Project's people-list entry for the Kurds (or IMB's, ROP3's, AX's, Etnopedia's, or another scheme's) — a record with its scheme's code, identity criteria, and a prefLabel like "Kurds." Lives under dul:Concept + skos:Concept. The concept has a name, a code, and criteria — not a population, a place, or a reachedness percentage. Those belong to the people the concept names.
gc:PeopleGroupCollective — the Kurds themselves. The actual people the classification concept names, considered across all places and times. The broader who. Carries gc:classifiedByPeopleGroup back to the concept.
gc:PeopleGroupPopulationSegment — a scoped slice of a collective. The reusable A-box anchor for population estimates, reachedness assessments, and engagement attestations. One collective is realized by many segments simultaneously: a cross-country rollup, one in-country slice per country, one in-city slice per city the group is reported in, one in-diaspora slice per host place.
gc:PeopleGroupCommunity — a local social manifestation. The community on the ground — shared identity, practice, and interaction patterns. A neighborhood, a tribal village, a diaspora hub. One segment can host many communities; one community can sit within multiple segments.

Why split the classification concept from the people

A classification concept is a record in a registry; the people are the actual humans. The JP entry titled "Kurds" carries a ROP3 code, identity criteria, and a prefLabel — it has no population, lives in no country, and isn't more or less reached. Asking "what is the Kurds' population?" is meaningful only when scoped: in Iraq? worldwide? in the German diaspora? Each answer attaches to a different gc:PeopleGroupPopulationSegment — a scoped slice of the Kurdish collective — not to the classification concept itself.

The shape resolves several problems at once. Multi-source disagreement is preserved (two schemes can classify the same segment under different concepts; both classifications coexist with their own provenance chains). Multi-source population numbers are preserved (JP and a local field-research team can each emit their own gc:PeopleGroupPopulationEstimate on the same segment). Sub-country and operational scopes (in-City, in-Diaspora, in-Church, in-MinistryEngagement) compose at query time without minting new subclasses. And the Lausanne maximality definition stays attached to the concept, where it belongs.

The four-layer model

Layer	Class	What it represents	Identity
Classification concept	`gc:PeopleGroup`	The publishing scheme's entry that names a people group — JP / IMB / ROP3 / AX's record with its code, identity criteria, and a prefLabel like "Kurds"	Constituted by the governing scheme's definition
Abstract collective	`gc:PeopleGroupCollective`	The actual people the concept names, considered across all places and times	One per concept
Scoped slice	`gc:PeopleGroupPopulationSegment`	A specific scope of the collective — across countries, in country, in city, in diaspora, in language, in religion, in church, in network, in denomination, in ministry engagement	One per (concept × scope dimension)
Local body	`gc:PeopleGroupCommunity`	A community on the ground — neighborhood, village, diaspora hub	Many per segment; one community can sit within multiple segments

Estimates and assessments hang off the segment, not the concept and not the community. Engagement attestations target a segment (gc:claimedPopulationSegment). Cross-source classifications target a segment (or a community). The concept stays clean: a name plus a gc:hasPeopleClassification link to the corresponding HIS ROP3 SKOS concept that serves as the cross-source join key.

Scope types

gc:PeopleGroupScopeTypeScheme carries 17 controlled-vocabulary concepts that classify the scope dimension of a gc:PeopleGroupPopulationSegment. Every segment must declare exactly one scope type via gc:hasScopeType. The corresponding scope dimension is carried via gc:locatedIn, gc:spatialScope, gc:hostPlace / gc:homelandPlace, gc:withinChurch, gc:withinNetwork, gc:withinDenomination, gc:withinMinistryEngagement, or gc:temporalScope.

Cross-source rollups

Code	Concept	When to use it
`PGST-PGA`	People Group Across Countries (PGAC)	The cross-country rollup of a concept — the segment that spans every country in which the group is reported. The natural carrier for global counts and global reachedness rollups.
`PGST-PGI`	People Group In Country (PGIC)	The per-country segment of a concept — one segment per (concept × ROG country) pair known to the source. The canonical join target for engagement attestations and country-scoped reachedness assessments.

Geographic finer-grained scopes

Code	Concept	When to use it
`PGST-CIT`	People Group In City	A segment scoped to a specific city. Useful for urban-targeted ministry, urban diaspora, or city-level Hotspot lists.
`PGST-ADM`	People Group In Admin Area	A segment scoped to an administrative subdivision below the country level (state, province, governorate, ADM1 / ADM2). The carrier for sub-national Hotspot lists.
`PGST-RGN`	People Group In Region	A segment scoped to a multi-country region (HIS ROG region, IMB affinity-region grouping).
`PGST-PLY`	People Group In Polygon	A segment scoped to an arbitrary polygon — a watershed, a tribal territory, a refugee-camp zone — that does not align with administrative geography.
`PGST-PLC`	People Group In Place	A segment scoped to a named place that is not a city, admin area, or region — a refugee camp, an island, a settlement.

Demographic scopes

Code	Concept	When to use it
`PGST-LNG`	People Group In Language	A segment scoped to a specific language community within a multilingual concept.
`PGST-REL`	People Group In Religion	A segment scoped to a religious sub-population — e.g., the Sunni Kurdish segment vs. the Yezidi Kurdish segment.
`PGST-CCT`	People Group In Caste, Clan, or Tribe	A segment scoped to a constituent caste, clan, or tribe within an aggregating concept.
`PGST-AFG`	People Group In Affinity Group	A segment scoped to an IMB-style affinity grouping — kinship, professional, or shared-experience cohort.
`PGST-CLU`	People Group In Cluster	A segment scoped to a cluster of related people groups under a higher-level grouping (HIS People Cluster, IMB Kinship Group).

Diaspora

Code	Concept	When to use it
`PGST-DIA`	People Group In Diaspora	A segment scoped to a diaspora population — persons of the people group living outside the homeland. Carries `gc:hostPlace` and `gc:homelandPlace`.

Operational scopes

Code	Concept	When to use it
`PGST-MEG`	People Group In Ministry Engagement	A segment scoped to the population reached by a specific engagement (`gc:EngagementSituation`, `gc:EngagementActivity`, or `gc:MinistryActivity`). Carries `gc:withinMinistryEngagement`.
`PGST-CHU`	People Group In Church	A segment scoped to the population gathered within a specific church or ekklesia body. Carries `gc:withinChurch`.
`PGST-NET`	People Group In Network	A segment scoped to the population organized within a specific network of churches or ministries. Carries `gc:withinNetwork`.
`PGST-DEN`	People Group In Denomination	A segment scoped to the population gathered within a specific denomination. Carries `gc:withinDenomination`.

Segments compose hierarchically: a PGAC segment may contain per-country PGIC segments via gc:hasSubPopulationSegment; a PGIC segment may contain per-city or per-cluster sub-segments. Non-hierarchical overlap (a Kurds-in-Iraq PGIC slice overlapping a Kurds-Diaspora-in-Germany segment via the moving population) is recorded with gc:overlapsPopulationSegment.

Classification: placing a concept under one or more schemes

A gc:PeopleGroupClassification is a time-indexed source-attributed claim that a particular gc:PeopleGroup concept classifies a particular entity (typically a gc:PeopleGroupPopulationSegment or gc:PeopleGroupCommunity). gc:PeopleGroupClassificationScheme instances are minted per source — one for the Joshua Project people-list, one for the HIS ROP3 registry, one for the IMB Affinity Bloc framework, one for AX, one for Etnopedia, one for a local research team's naming scheme.

Classification is its own first-class entity rather than a property on the segment, which preserves disagreement: when JP's ROP3-anchored scheme and a local-research team classify the same segment under different concepts (one publishes a single "Kurds" concept; another splits it into "Kurmanji" and "Sorani" sub-groupings), both classifications coexist with their own provenance chains. Cross-scheme alignment runs through skos:exactMatch / skos:closeMatch on the contained concepts, with ROP3 typically serving as the join key.

Code
 
<https://data.global.church/peoplegroup/classification/kurd-IQ-jp>
  a gc:PeopleGroupClassification ;
  gc:classificationScheme  <https://data.global.church/peoplegroup/scheme/joshua-project> ;
  gc:classificationConcept <https://data.global.church/peoplegroup/concept/kurd> ;
  gc:classifiedEntity      <https://data.global.church/segments/kurd/IQ> ;
  prov:wasAttributedTo     <https://data.global.church/org/joshua-project> ;
  prov:generatedAtTime     "2026-01-15T10:00:00Z"^^xsd:dateTime .

Population estimates: multi-source disagreement preserved

A gc:PeopleGroupPopulationEstimate is an information-bearing datum about a segment — count, percent Christian, percent evangelical, source metric, confidence, source, date. It's a gc:AssessmentDatum so it inherits the finer-grained provenance machinery (source, confidence, generation time) that the assessment family uses.

Multi-source disagreement is preserved by minting separate estimates per source rather than coalescing values onto the segment. JP and a local field-research team can each emit their own estimate for Kurds-in-Iraq; both attach to the same segment via gc:hasPopulationEstimate. If anyone authors a reconciled estimate later, that is a third entity that prov:wasDerivedFrom both sources — not an overwrite.

Code
 
<https://data.global.church/segments/kurd/IQ>
  a gc:PeopleGroupPopulationSegment ;
  gc:hasPopulationEstimate
    <https://data.global.church/peoplegroup/estimate/kurd-IQ-jp-2026> ,
    <https://data.global.church/peoplegroup/estimate/kurd-IQ-local-2026> .

<https://data.global.church/peoplegroup/estimate/kurd-IQ-jp-2026>
  a gc:PeopleGroupPopulationEstimate ;
  gc:estimateForPopulationSegment <https://data.global.church/segments/kurd/IQ> ;
  gc:datumValue        "6500000"^^xsd:integer ;
  gc:confidenceScore   "0.85"^^xsd:decimal ;
  prov:wasAttributedTo <https://data.global.church/org/joshua-project> ;
  prov:generatedAtTime "2026-01-15T10:00:00Z"^^xsd:dateTime .

<https://data.global.church/peoplegroup/estimate/kurd-IQ-local-2026>
  a gc:PeopleGroupPopulationEstimate ;
  gc:estimateForPopulationSegment <https://data.global.church/segments/kurd/IQ> ;
  gc:datumValue        "7100000"^^xsd:integer ;
  gc:confidenceScore   "0.70"^^xsd:decimal ;
  prov:wasAttributedTo <https://data.global.church/org/example-field-team> ;
  prov:generatedAtTime "2026-03-01T14:00:00Z"^^xsd:dateTime .

Reachedness assessment results

A gc:ReachednessAssessmentResult is the source-specific assessment-result of gospel access, evangelical presence, church presence, and engagement on a gc:PeopleGroupPopulationSegment. It's a gc:AssessmentResult, so it slots into the methodology + result-target machinery shared with people-group and church-health assessments — see Assessment Frameworks.

A reachedness result links back to its segment via gc:resultForPopulationSegment (or the parallel inverse gc:assessmentResultOfPopulationSegment) and carries one or more reachedness payloads:

Property	Range	What it carries
`gc:hasReachednessStatus`	`skos:Concept`	The reachedness status reported by this result — typically a JP / IMB / AX scheme value (e.g., a JP Progress Scale concept).
`gc:hasEngagementStatus`	`skos:Concept`	The engagement status — typically an IMB engagement-status value or a `poe:PhasesOfEngagementScheme` phase.
`gc:percentEvangelical`	`xsd:double`	Estimated percentage of the segment's population that is evangelical Christian (0.0–100.0).
`gc:percentChristian`	`xsd:double`	Estimated percentage that is Christian-adherent (0.0–100.0).

The result is the natural carrier for verification level (V1–V4) and data confidence (Low/Med/High) when those are scoped per-segment rather than per-claim. As with population estimates, multi-source disagreement is preserved: JP, IMB, AX, Etnopedia, and a local research team can each emit their own reachedness result on the same segment, and downstream tools roll up per-source rather than coercing cross-framework equivalences.

Code
 
<https://data.global.church/peoplegroup/reachedness/kurd-IQ-jp-2026>
  a gc:ReachednessAssessmentResult ;
  gc:resultForPopulationSegment <https://data.global.church/segments/kurd/IQ> ;
  gc:hasReachednessStatus jp:JPScale1 ;
  gc:percentEvangelical "0.1"^^xsd:double ;
  gc:percentChristian   "0.3"^^xsd:double ;
  gc:resultDate         "2026-01-15T10:00:00Z"^^xsd:dateTime ;
  prov:wasAttributedTo  <https://data.global.church/org/joshua-project> ;
  prov:generatedAtTime  "2026-01-15T10:00:00Z"^^xsd:dateTime .

Multi-tier assessment hooks

The same Assessment / AssessmentResult infrastructure that scores people groups, churches, and disciples extends naturally to per-segment assessment via two subproperties of the union-range gc:assessmentTarget / gc:resultTarget superproperties:

Subproperty	Of	Range
`gc:assessesPopulationSegment`	`gc:assessmentTarget`	`gc:PeopleGroupPopulationSegment`
`gc:resultForPopulationSegment`	`gc:resultTarget`	`gc:PeopleGroupPopulationSegment`

Existing concept-targeted data (gc:assessesPeopleGroup / gc:resultForPeopleGroup) continues to validate against the broader superproperty without re-emission. New per-segment assessments use the narrower subproperties and are reachable from queries that walk gc:assessmentTarget regardless of which tier they were authored at. See Assessment Frameworks for the methodology side.

Querying the People Group model

Find a segment, its concept, and its latest population estimate

Code
 
PREFIX gc:   <https://ontology.global.church/core#>
PREFIX prov: <http://www.w3.org/ns/prov#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT ?segment ?conceptLabel ?value ?source ?asOf WHERE {
  ?segment a gc:PeopleGroupPopulationSegment ;
           gc:ofPeopleGroup ?concept ;
           gc:hasPopulationEstimate ?estimate .
  ?concept skos:prefLabel ?conceptLabel .
  ?estimate gc:datumValue ?value ;
            prov:wasAttributedTo ?source ;
            prov:generatedAtTime ?asOf .
}
ORDER BY ?segment DESC(?asOf)

Find all per-country segments inside a region

Code
 
PREFIX gc:   <https://ontology.global.church/core#>
PREFIX rog:  <https://data.global.church/his/rog/>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

SELECT ?segment ?country WHERE {
  ?segment a gc:PeopleGroupPopulationSegment ;
           gc:hasScopeType gc:ScopeTypePGIC ;
           gc:locatedIn ?country .
  ?country skos:broader rog:region-africa .
}

Replace gc:ScopeTypePGIC with gc:ScopeTypePGAC to find the cross-country rollup segments instead, or with gc:ScopeTypePGInDiaspora for diaspora-scoped segments.

Find segments where two or more sources disagree on population

Code
 
PREFIX gc:   <https://ontology.global.church/core#>
PREFIX prov: <http://www.w3.org/ns/prov#>

SELECT ?segment (COUNT(DISTINCT ?source) AS ?nSources)
              (MIN(?value) AS ?lowEstimate)
              (MAX(?value) AS ?highEstimate)
WHERE {
  ?segment gc:hasPopulationEstimate ?estimate .
  ?estimate gc:datumValue ?value ;
            prov:wasAttributedTo ?source .
}
GROUP BY ?segment
HAVING (COUNT(DISTINCT ?source) >= 2)
ORDER BY DESC(?highEstimate - ?lowEstimate)

This is the kind of query the multi-source shape unlocks: which segments have the largest source-to-source disagreement on population? — without a normalization step that hides the disagreement.

Where the schemes live in the ontology

Scheme	URI
`gc:PeopleGroupScopeTypeScheme`	`https://ontology.global.church/core#PeopleGroupScopeTypeScheme`
`gc:PeopleGroupClassificationScheme` (per-source instances)	minted per scheme, e.g. `https://data.global.church/peoplegroup/scheme/joshua-project`

Full per-concept definitions, scope notes, and relationships are published at the Ontology Reference under SKOS Vocabularies.