kempersc c88fd3af70 Refactor code structure for improved readability and maintainability

2026-01-09 11:05:26 +01:00

16 KiB

Raw Blame History

PiCo Ontology Analysis

Version: 0.1.0
Last Updated: 2025-01-09
Source: https://github.com/CBG-Centrum-voor-familiegeschiedenis/PiCo
Related: Executive Summary | Claims and Provenance

1. Overview

PiCo (Persons in Context) is an ontology developed by the CBG-Centrum-voor-familiegeschiedenis (Center for Family History) in the Netherlands. It provides a conceptual framework for modeling persons in historical sources with explicit distinction between observations (what sources say) and reconstructions (what we conclude).

This distinction is fundamental to the PPID design and directly informs our two-level identifier architecture.

2. Core Philosophy

2.1 The Observation-Reconstruction Distinction

PiCo's central innovation is the explicit separation of:

Concept	Definition	Example
PersonObservation	A person as described in a specific source	"The baptism register states 'Johannes, son of Pieter'"
PersonReconstruction	A curated identity derived from one or more observations	"Johannes Pietersen van der Berg (1692-1756)"

This mirrors the genealogical research process:

┌─────────────────────────────────────────────────────────────────┐
│                     RESEARCH WORKFLOW                            │
│                                                                  │
│  Source A          Source B          Source C                   │
│  (Baptism)         (Marriage)        (Burial)                   │
│      │                 │                 │                       │
│      ▼                 ▼                 ▼                       │
│  ┌────────┐       ┌────────┐       ┌────────┐                   │
│  │ Person │       │ Person │       │ Person │                   │
│  │ Obs. A │       │ Obs. B │       │ Obs. C │                   │
│  └────────┘       └────────┘       └────────┘                   │
│      │                 │                 │                       │
│      └────────────────┬┴─────────────────┘                       │
│                       │                                          │
│                       ▼  (researcher reasoning)                  │
│              ┌─────────────────┐                                 │
│              │     Person      │                                 │
│              │ Reconstruction  │                                 │
│              │ "Johannes..."   │                                 │
│              └─────────────────┘                                 │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

2.2 Why This Matters

Benefit	Description
Transparency	Clear separation of evidence from conclusions
Traceability	Every assertion traceable to source
Revision safety	New evidence can update reconstruction without losing observations
Scholarly integrity	Supports genealogical proof standards
Conflict handling	Contradictory sources can coexist

3. Ontology Structure

3.1 Namespace and Prefixes

@prefix picom: <https://personsincontext.org/model#> .
@prefix pnv:   <https://w3id.org/pnv#> .
@prefix prov:  <http://www.w3.org/ns/prov#> .
@prefix bio:   <http://purl.org/vocab/bio/0.1/> .
@prefix schema: <http://schema.org/> .

3.2 Class Hierarchy

schema:Person
    │
    ├── picom:PersonObservation
    │       │
    │       └── (represents a person as found in a single source)
    │
    └── picom:PersonReconstruction
            │
            └── (represents a curated person identity)

3.3 Core Classes

PersonObservation

picom:PersonObservation a owl:Class ;
    rdfs:subClassOf schema:Person ;
    rdfs:label "Person Observation"@en ;
    rdfs:comment """A person as observed/described in a specific source. 
                    This represents what the source says, not necessarily 
                    what is true."""@en .

Key properties:

picom:hasName → Name as recorded in source
picom:hasRole → Role mentioned in source
picom:inRecord → Link to source document
prov:wasDerivedFrom → Source provenance

PersonReconstruction

picom:PersonReconstruction a owl:Class ;
    rdfs:subClassOf schema:Person ;
    rdfs:label "Person Reconstruction"@en ;
    rdfs:comment """A curated person identity constructed from one or more 
                    PersonObservations through research and reasoning."""@en .

Key properties:

prov:wasDerivedFrom → Links to source PersonObservations
picom:hasName → Canonical name form(s)
bio:birth / bio:death → Life events
picom:hasRole → Aggregated roles

4. Integration with Existing Ontologies

PiCo builds on established vocabularies rather than reinventing:

4.1 Schema.org

PiCo Usage	Schema.org Class/Property
Person base class	`schema:Person`
Birth date	`schema:birthDate`
Death date	`schema:deathDate`
Gender	`schema:gender`
Family name	`schema:familyName`
Given name	`schema:givenName`

4.2 PROV-O (Provenance Ontology)

PiCo Usage	PROV-O Property
Observation derived from source	`prov:wasDerivedFrom`
Reconstruction generated by activity	`prov:wasGeneratedBy`
Attribution to researcher	`prov:wasAttributedTo`
Revision tracking	`prov:wasRevisionOf`

# Example: Reconstruction with provenance
<reconstruction/johannes-van-der-berg>
    a picom:PersonReconstruction ;
    prov:wasDerivedFrom <observation/baptism-1692-123> ;
    prov:wasDerivedFrom <observation/marriage-1715-456> ;
    prov:wasDerivedFrom <observation/burial-1756-789> ;
    prov:wasGeneratedBy <research-activity/cbg-2024-001> ;
    prov:wasAttributedTo <researcher/jan-jansen> .

4.3 BIO Vocabulary

PiCo Usage	BIO Class/Property
Birth event	`bio:Birth`
Death event	`bio:Death`
Marriage	`bio:Marriage`
Event date	`bio:date`
Event place	`bio:place`

4.4 PNV (Person Name Vocabulary)

PiCo uses PNV for structured name representation:

<observation/baptism-1692-123>
    picom:hasName [
        a pnv:PersonName ;
        pnv:givenName "Johannes" ;
        pnv:patronym "Pietersen" ;
        pnv:surnamePrefix "van der" ;
        pnv:baseSurname "Berg" ;
        pnv:literalName "Johannes Pietersen van der Berg"
    ] .

5. Person Name Vocabulary (PNV) Deep Dive

5.1 Background

PNV was developed to handle the complexity of Dutch historical names, but its patterns apply globally:

Patronymics: "Pietersen" (son of Pieter)
Surname prefixes: "van der", "de", "ten"
Multiple given names
Initials
Name changes over time

5.2 PNV Properties

Property	Description	Example
`pnv:literalName`	Full name as single string	"Johannes Pietersen van der Berg"
`pnv:givenName`	First/given name(s)	"Johannes"
`pnv:patronym`	Patronymic name	"Pietersen"
`pnv:surnamePrefix`	Particles before surname	"van der"
`pnv:baseSurname`	Core family name	"Berg"
`pnv:surname`	Combined prefix + baseSurname	"van der Berg"
`pnv:initials`	Initials only	"J.P."
`pnv:infixTitle`	Title within name	"graaf" (count)
`pnv:disambiguatingDescription`	Distinguishing info	"de oude" (the elder)

5.3 Name Complexity Examples

Dutch with patronymic:

[ a pnv:PersonName ;
  pnv:givenName "Jan" ;
  pnv:patronym "Hendrikszoon" ;
  pnv:surnamePrefix "van" ;
  pnv:baseSurname "Amstel" ;
  pnv:literalName "Jan Hendrikszoon van Amstel" ] .

Spanish with two family names:

[ a pnv:PersonName ;
  pnv:givenName "María" ;
  pnv:givenName "Elena" ;
  pnv:baseSurname "García" ;
  pnv:baseSurname "López" ;
  pnv:literalName "María Elena García López" ] .

Icelandic patronymic (no surname):

[ a pnv:PersonName ;
  pnv:givenName "Björk" ;
  pnv:patronym "Guðmundsdóttir" ;
  pnv:literalName "Björk Guðmundsdóttir" ] .

6. Handling Uncertainty

6.1 Date Uncertainty

PiCo allows flexibility in date representation:

# Exact date known
<observation/birth-1692>
    bio:date "1692-03-15"^^xsd:date .

# Only year known
<observation/birth-approx>
    bio:date "1692"^^xsd:gYear .

# Estimated from age at death
<observation/birth-estimated>
    picom:estimatedBirthYear "1692"^^xsd:gYear ;
    picom:birthYearEstimationMethod "calculated from age 64 at death in 1756" .

6.2 Uncertain Identity Linkage

When observations might refer to same person:

<observation/a> picom:possibleSameAs <observation/b> .
<observation/a> picom:certainSameAs <observation/c> .

6.3 Confidence Scores

PiCo supports confidence assertions:

<reconstruction/johannes>
    picom:hasConfidence [
        picom:confidenceValue 0.85 ;
        picom:confidenceMethod "probabilistic record linkage" ;
        picom:confidenceNote "High confidence based on matching name, date, and location"
    ] .

7. Role Modeling

7.1 Persons in Context

PiCo's name reflects its focus on persons in context - roles and relationships:

<observation/baptism-1692-123>
    picom:hasRole [
        a picom:Role ;
        picom:roleType "child" ;
        picom:roleContext <event/baptism-1692>
    ] ;
    picom:hasRole [
        a picom:Role ;
        picom:roleType "son" ;
        picom:roleInRelationTo <observation/pieter-father>
    ] .

7.2 Role Types for Heritage Sector

Role Type	Context	Example
`archivist`	Institution employment	"Chief archivist at Noord-Hollands Archief"
`curator`	Collection management	"Curator of Dutch Masters"
`director`	Leadership	"Museum director 2010-2020"
`donor`	Collection contribution	"Donated family papers in 1985"
`researcher`	Academic work	"Visiting researcher"
`subject`	Collection content	"Person depicted in portrait"

8. PPID Alignment with PiCo

8.1 Mapping PiCo to PPID

PiCo Concept	PPID Implementation
`picom:PersonObservation`	POID (Person Observation ID)
`picom:PersonReconstruction`	PRID (Person Reconstruction ID)
`prov:wasDerivedFrom`	Links PRID → POIDs
`pnv:PersonName`	Structured name storage
`picom:hasRole`	Role at heritage institution

8.2 Extended PPID Model

PPID extends PiCo for heritage custodian context:

@prefix ppid: <https://ppid.org/> .
@prefix picom: <https://personsincontext.org/model#> .
@prefix ghcid: <https://w3id.org/heritage/custodian/> .

# Person Observation (from LinkedIn)
ppid:POID-7a3b-c4d5-e6f7-8901 a picom:PersonObservation ;
    picom:hasName [
        pnv:givenName "Jan" ;
        pnv:baseSurname "Berg" ;
        pnv:literalName "Jan van den Berg"
    ] ;
    picom:hasRole [
        picom:roleType "Senior Archivist" ;
        picom:roleAtInstitution ghcid:NL-NH-HAA-A-NHA
    ] ;
    prov:wasDerivedFrom <https://linkedin.com/in/jan-van-den-berg> ;
    ppid:retrievedOn "2025-01-09"^^xsd:date .

# Person Observation (from institutional website)
ppid:POID-8b4c-d5e6-f7g8-9012 a picom:PersonObservation ;
    picom:hasName [
        pnv:givenName "J." ;
        pnv:surnamePrefix "van den" ;
        pnv:baseSurname "Berg" ;
        pnv:literalName "J. van den Berg"
    ] ;
    picom:hasRole [
        picom:roleType "Archivaris" ;
        picom:roleAtInstitution ghcid:NL-NH-HAA-A-NHA
    ] ;
    prov:wasDerivedFrom <https://noord-hollandsarchief.nl/over-ons/medewerkers> ;
    ppid:retrievedOn "2025-01-08"^^xsd:date .

# Person Reconstruction (curated identity)
ppid:PRID-1234-5678-90ab-cdef a picom:PersonReconstruction ;
    picom:hasName [
        pnv:givenName "Jan" ;
        pnv:surnamePrefix "van den" ;
        pnv:baseSurname "Berg" ;
        pnv:literalName "Jan van den Berg"
    ] ;
    prov:wasDerivedFrom ppid:POID-7a3b-c4d5-e6f7-8901 ;
    prov:wasDerivedFrom ppid:POID-8b4c-d5e6-f7g8-9012 ;
    prov:wasGeneratedBy [
        a prov:Activity ;
        prov:wasAssociatedWith <agent/ppid-matcher> ;
        prov:atTime "2025-01-09T10:30:00Z"^^xsd:dateTime
    ] ;
    ppid:employmentHistory [
        ppid:institution ghcid:NL-NH-HAA-A-NHA ;
        ppid:role "Senior Archivist" ;
        ppid:startDate "2015"^^xsd:gYear ;
        ppid:endDate "present"
    ] .

9. Implementation Considerations

9.1 When to Create POID vs PRID

Scenario	Create
Extract person from LinkedIn	POID
Extract person from institutional website	POID
Extract person from archival document	POID
Match multiple POIDs to single identity	PRID
User claims "these are the same person"	PRID linking POIDs

9.2 PRID Creation Rules

A PRID should be created when:

Single authoritative source: One high-quality POID with comprehensive data
Multiple matched POIDs: Algorithm or human determines multiple observations refer to same person
External identifier exists: Person has ORCID, ISNI, or Wikidata ID

9.3 Handling Updates

# Original reconstruction
ppid:PRID-1234-5678-90ab-cdef a picom:PersonReconstruction ;
    prov:wasGeneratedAt "2025-01-09T10:30:00Z"^^xsd:dateTime .

# Updated reconstruction (new evidence)
ppid:PRID-1234-5678-90ab-cdef-v2 a picom:PersonReconstruction ;
    prov:wasRevisionOf ppid:PRID-1234-5678-90ab-cdef ;
    prov:wasDerivedFrom ppid:POID-7a3b-c4d5-e6f7-8901 ;
    prov:wasDerivedFrom ppid:POID-8b4c-d5e6-f7g8-9012 ;
    prov:wasDerivedFrom ppid:POID-new-observation ;  # New evidence
    prov:wasGeneratedAt "2025-01-15T14:00:00Z"^^xsd:dateTime .

10. Gaps in PiCo for PPID

While PiCo provides an excellent foundation, PPID needs extensions:

Gap	PPID Extension
Web source provenance	Add XPath, retrieval timestamp, HTML archival
Confidence scoring standards	Define confidence scale and methods
Heritage sector roles	Vocabulary for archivist, curator, director, etc.
Institution linking	Integration with GHCID
Living person data protection	GDPR-compliant access controls

These extensions are detailed in 07_claims_and_provenance.md and 08_implementation_guidelines.md.

11. References

Primary Sources

PiCo Ontology: https://github.com/CBG-Centrum-voor-familiegeschiedenis/PiCo
PiCo Documentation: https://personsincontext.org/
PNV Specification: https://w3id.org/pnv

Schema.org Person: https://schema.org/Person
BIO Vocabulary: http://purl.org/vocab/bio/0.1/
PROV-O: https://www.w3.org/TR/prov-o/

Academic Papers

Bloothooft, G., & Schraagen, M. (2015). "Learning name variants from true person resolution." Proceedings of the First Workshop on Computational Models of Reference, Anaphora and Coreference.

16 KiB Raw Blame History