Genealogical Relationships Research

Why This Matters

Genealogical systems have decades of experience modeling:

Billions of person records (FamilySearch has 1.5+ billion)
Complex relationships: biological, adoptive, step, foster, guardianship
Temporal relationships: marriages, divorces, remarriages
Evidence and certainty: confidence levels, source citations, proof standards
Non-traditional families: same-sex couples, blended families

They’ve solved hard problems in relationship modeling that apply to any entity graph.

Key Standards

Standard	Era	Status	Purpose
GEDCOM 5.5	1995	Legacy (widely used)	Text-based file format for family tree exchange
GEDCOM 5.5.1	1999	Industry standard	Minor update, still dominant
GEDCOM X	2012	Active	Modern data model for evidence-based genealogy
FamilySearch GEDCOM 7.0	2021	Current	Updated file format building on GEDCOM X concepts

Core Relationship Model

FamilySearch’s Two Relationship Types

FamilySearch’s Family Tree API supports exactly two relationship types:

Relationship Type	Description	Structure
Couple Relationship	A relationship between two people (spouses/partners)	Links two `Person` records
Child-and-Parents Relationship	A relationship between a child and their parents	Links child to up to two parents

This is remarkably simple. All family structures emerge from just these two primitives.

GEDCOM’s Family-Centric Model

Traditional GEDCOM uses a FAM (Family) record as the linking structure:

0 @F1@ FAM
1 HUSB @I1@        (husband/partner 1)
1 WIFE @I2@        (wife/partner 2)
1 CHIL @I3@        (child)
1 CHIL @I4@        (another child)
1 MARR
2 DATE 15 JUN 1985
2 PLAC Chicago, Illinois

The FAM record is the only source of links between individuals. This is similar to a join table in relational databases.

Key Insight: Sibling Relationships Are Implied

Neither GEDCOM nor FamilySearch explicitly models sibling relationships. Instead:

Siblings are inferred from shared parents
Half-siblings share exactly one parent
Step-siblings share no biological parents but have parents in a couple relationship

This avoids redundancy and keeps the model simple.

Parent-Child Relationship Types

FamilySearch Parent-Child Facts

Type URI	Description
`http://gedcomx.org/BiologicalParent`	Birth parent
`http://gedcomx.org/AdoptiveParent`	Legally adopted the child
`http://gedcomx.org/StepParent`	Married to biological parent
`http://gedcomx.org/FosterParent`	Temporary guardian
`http://gedcomx.org/GuardianParent`	Legal guardian

GEDCOM PEDI (Pedigree) Tag

GEDCOM 5.5.1 uses the PEDI tag to specify parent-child relationship type:

0 @I1@ INDI
1 NAME Child /Name/
1 FAMC @F1@
2 PEDI birth          (biological)
1 FAMC @F2@
2 PEDI adopted        (adoptive parents)

PEDI values:

birth — biological relationship
adopted — legal adoption
foster — foster care
sealing — LDS temple sealing (FamilySearch-specific)

Multiple Parents

A person can have multiple sets of parents:

Biological parents
Adoptive parents
Step-parents
Foster parents

Each is a separate Child-and-Parents relationship with its own type qualifier.

Couple Relationship Facts

Supported Couple Events

Type URI	Description	Date	Place	Multiple
`http://gedcomx.org/Marriage`	Legal marriage	Yes	Yes	Yes
`http://gedcomx.org/Divorce`	Legal divorce	Yes	Yes	Yes
`http://gedcomx.org/Annulment`	Marriage annulment	Yes	Yes	Yes
`http://gedcomx.org/CommonLawMarriage`	Common law union	Yes	Yes	Yes
`http://familysearch.org/v1/LivedTogether`	Cohabitation	No	No	No
`http://familysearch.org/v1/CoupleNeverHadChildren`	Flag for childless couple	No	No	No

Temporal Relationship Handling

Marriage as an Event: Marriage in GEDCOM is an event with date and place, not a boolean status. The MARR record documents when/where the marriage occurred.

Multiple Marriages: When couples marry, divorce, and remarry (even to the same person), best practice is:

Create a new FAM record for each marriage
Don’t combine multiple MARR/DIV events in one record
This maintains chronological clarity and allows other relationships in between

Divorce Without Marriage Record: GEDCOM allows a DIV (divorce) record without a corresponding MARR if the marriage date is unknown but divorce is documented.

Evidence and Confidence

GEDCOM X Conclusion Model

GEDCOM X models genealogical data as conclusions that can have:

Property	Type	Purpose
`confidence`	ConfidenceLevel	How certain is this data?
`source`	SourceReference[]	What sources support this?
`attribution`	Attribution	Who contributed this? When?
`analysis`	Document reference	Written reasoning/proof
`note`	Note[]	Additional context

Every fact, relationship, and person in GEDCOM X is a “conclusion” that can carry this metadata.

Genealogical Proof Standard (GPS)

The GPS is the established framework for genealogical certainty:

Reasonably exhaustive research in reliable sources
Complete and accurate source citations for all facts
Analysis and correlation of collected information
Resolution of conflicting evidence
Soundly reasoned, coherently written conclusion

Meeting GPS doesn’t mean “beyond doubt” — it’s closer to legal “clear and convincing” standard. Conclusions can be revisited when new evidence emerges.

Evidence Types

Type	Definition	Example
Direct Evidence	Directly answers the question	Birth certificate listing parents
Indirect Evidence	Requires interpretation/correlation	Census showing household composition
Negative Evidence	Absence suggests something	Person not in death records = likely alive

Source Classification

Source Type	Definition	Reliability
Original	Created at/near time of event by firsthand observer	Higher
Derivative	Copied, transcribed, or abstracted from original	Lower
Authored	Created after the fact (family histories, biographies)	Varies

GEDCOM X vs GEDCOM 5.5

Fundamental Philosophy Shift

Aspect	GEDCOM 5.5	GEDCOM X
Focus	Conclusions only (the family tree)	Evidence + conclusions (the research process)
Sources	Optional attachments	First-class citizens
Confidence	Not supported	Built into conclusion model
Format	Text-based, single format	JSON/XML, multiple serializations
License	Proprietary (LDS)	Open source (Apache 2.0)
Media	External links only	Bundled in file format

What GEDCOM X Added

Source Descriptions — Rich metadata about sources
Source References — Link conclusions to supporting sources
Evidence References — Explicitly model evidence chains
Contributor Attribution — Track who added what and when
Analysis Documents — Written reasoning and proof arguments
Confidence Levels — Express certainty about conclusions

Evolution Timeline

1984: GEDCOM created by LDS Church
1995: GEDCOM 5.5 released
1999: GEDCOM 5.5.1 (minor update)
2012: GEDCOM X introduced (evidence-based model)
2021: FamilySearch GEDCOM 7.0 (modernized file format)

Complex Family Situations

Adoption Modeling

GEDCOM approach:

0 @I1@ INDI
1 NAME Adopted /Child/
1 FAMC @F1@           (biological family)
2 PEDI birth
1 FAMC @F2@           (adoptive family)
2 PEDI adopted
1 ADOP                (adoption event)
2 DATE 11 JAN 1990
2 FAMC @F2@           (which family adopted)
3 ADOP BOTH           (both parents adopted)

FamilySearch approach:

Create Child-and-Parents relationship to biological parents with BiologicalParent fact
Create separate Child-and-Parents relationship to adoptive parents with AdoptiveParent fact
Both relationships coexist

Step-Parent Families

When a biological parent remarries:

Child has BiologicalParent relationship to birth parents
Child can have StepParent relationship to parent’s new spouse
No need to modify biological relationships

Blended Families (“Patchwork Families”)

Multiple FAMC (family child) links handle complex blended families:

Each child links to their biological parents
Additional links for step/adoptive relationships
Couple relationships tie adults together

Same-Sex Relationships

FamilySearch (as of recent update):

Allows documenting same-sex marriages and adoptions
Spouse selection no longer restricted by sex
Photos, stories, documents accepted for same-sex relationships
Required significant technical redesign of tree search systems

Limitations:

FamilySearch is operated by LDS Church
Same-sex couples cannot be sealed in temple
Children cannot be sealed to same-sex parents
This is genealogical documentation, not religious endorsement

GEDCOM 7.0:

Updated to allow same-sex partnerships
HUSB/WIFE terminology being reconsidered for neutrality

Limitations and Criticisms

GEDCOM 5.5 Problems

Issue	Description
Nuclear family assumption	Original design assumed husband/wife/children model
No embedded media	Only links to external files
Inconsistent implementations	Software interprets spec differently
Proprietary extensions	Vendors add non-standard tags
Slow evolution	26 years between 5.5.1 and 7.0
Contradictions in spec	Some requirements are internally inconsistent

The Compliance Paradox

Developers face a dilemma:

Comply strictly with GEDCOM spec → incompatible with common software
Match common practice → deviate from official standard

Many genealogy programs don’t fully implement the standard, creating an ecosystem of partial compatibility.

Family-Centric Model Limitations

The FAM record assumes:

Exactly 0-2 adults (HUSB/WIFE)
Any number of children
Adults are in a “couple” relationship

This struggles with:

Polyamorous relationships
Communal child-rearing
Cultures with different family structures
Situations where “family” has different meaning

Evidence Model Adoption

While GEDCOM X has excellent evidence modeling, many users:

Don’t use confidence levels
Skip source citations
Treat it like GEDCOM 5.5 (conclusions only)

The tools support sophisticated research; user behavior lags.

Data Structures Reference

GEDCOM 5.5.1 Individual Record

0 @I1@ INDI
1 NAME John /Smith/
1 SEX M
1 BIRT
2 DATE 1 JAN 1900
2 PLAC New York, New York
1 DEAT
2 DATE 15 MAR 1975
2 PLAC Los Angeles, California
1 FAMC @F1@           (family as child)
2 PEDI birth
1 FAMS @F2@           (family as spouse)

GEDCOM 5.5.1 Family Record

0 @F1@ FAM
1 HUSB @I2@
1 WIFE @I3@
1 CHIL @I1@
1 MARR
2 DATE 15 JUN 1985
2 PLAC Chicago, Illinois
1 DIV
2 DATE 3 MAR 1995

GEDCOM X Relationship (JSON)

{
  "id": "R1",
  "type": "http://gedcomx.org/Couple",
  "person1": { "resource": "#P1" },
  "person2": { "resource": "#P2" },
  "facts": [
    {
      "type": "http://gedcomx.org/Marriage",
      "date": { "original": "15 June 1985" },
      "place": { "original": "Chicago, Illinois" },
      "confidence": "http://gedcomx.org/High"
    }
  ],
  "sources": [
    { "description": "#S1" }
  ]
}

GEDCOM X Parent-Child Relationship (JSON)

{
  "id": "R2",
  "type": "http://gedcomx.org/ParentChild",
  "person1": { "resource": "#P1" },  // parent
  "person2": { "resource": "#P3" },  // child
  "facts": [
    {
      "type": "http://gedcomx.org/BiologicalParent"
    }
  ]
}

Lessons for Entity Graphs

1. Two Relationship Types Are Enough

FamilySearch models all family structures with just:

Couple relationships
Parent-child relationships

Sibling, grandparent, cousin, etc. are computed, not stored.

2. Relationships Are Entities, Not Edges

Relationships in GEDCOM X have:

Their own IDs
Facts/events attached
Source citations
Confidence levels
Temporal data

They’re full entities, not just foreign keys.

3. Evidence Matters

Sophisticated systems separate:

Information — raw data from sources
Evidence — information that answers a question
Conclusion — accepted answer with supporting evidence

This three-layer model enables proper reasoning and revision.

4. Temporal Relationships Need Events

Don’t model “married: true/false”. Model:

Marriage event (date, place)
Divorce event (date, place)
Remarriage event (date, place)

The current status is computed from event history.

5. Multiple Relationship Instances

A person can have multiple:

Parent relationships (biological + adoptive + step)
Spouse relationships (sequential marriages)
The same two people can have multiple relationships (married, divorced, remarried)

6. Types Are Extensible

Both GEDCOM and GEDCOM X support:

Standard relationship types
Custom types via data URIs
Extension mechanisms for domain-specific needs

Key Resources

Specifications

Developer Documentation

Research Standards

Genealogical Proof Standard
Evidence Explained by Elizabeth Shown Mills
Mastering Genealogical Proof by Thomas W. Jones
Genealogy Standards from Board for Certification of Genealogists

Community Resources

GEDCOM.io — Official GEDCOM specification site
GEDCOM X — GEDCOM X documentation
Tamura Jones GEDCOM Articles — Excellent technical analysis

Open Questions

How do other cultures model family? — GEDCOM assumes Western nuclear family. What about extended family structures, clan systems, or cultures with different kinship terminology?
What’s the right abstraction level? — FamilySearch’s two-relationship model is elegant but requires computation for common queries (siblings, grandparents). Is this the right trade-off?
How should confidence propagate? — If a parent-child relationship has low confidence, how does that affect inferred relationships (siblings, grandparents)?
Relationship vs. Role? — Is “biological parent” a relationship type or a role within a parent-child relationship? GEDCOM X uses “facts” on relationships, which is role-like.
Time modeling — Relationships have beginnings (marriage) and endings (divorce, death). How granular should temporal modeling be?