Files
thinkloom-openai-hackathon/schemas/provenance/v1/human-authorship-record.schema.json
T

589 lines
15 KiB
JSON

{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"$id": "https://thinkloom.app/schemas/provenance/1.0/human-authorship-record.schema.json",
"title": "Human Authorship Record of Provenance",
"description": "Canonical machine-readable, non-authoritative, exact-deposit HARP projection. Exact status is evidentiary only and cannot determine legal authorship, originality, copyrightability, or a human percentage. Its digest identity excludes harp_sha256.",
"type": "object",
"properties": {
"schema_version": {
"const": "1.0"
},
"harp_id": {
"type": "string",
"pattern": "^harp_[0-9A-HJKMNP-TV-Z]{26}$"
},
"project_id": {
"type": "string",
"pattern": "^project_[0-9A-HJKMNP-TV-Z]{26}$"
},
"application_version": {
"type": "string",
"pattern": "^\\d+\\.\\d+\\.\\d+$"
},
"generator_version": {
"type": "string",
"pattern": "^\\d+\\.\\d+\\.\\d+$"
},
"deposit": {
"type": "object",
"properties": {
"deposit_id": {
"type": "string",
"pattern": "^deposit_[0-9A-HJKMNP-TV-Z]{26}$"
},
"deposit_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
},
"manuscript_revision_id": {
"type": "string",
"pattern": "^revision_[0-9A-HJKMNP-TV-Z]{26}$"
},
"manuscript_revision_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
}
},
"required": [
"deposit_id",
"deposit_sha256",
"manuscript_revision_id",
"manuscript_revision_sha256"
],
"additionalProperties": false
},
"cpl_binding": {
"type": "object",
"properties": {
"chain_head": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
},
"event_sequence": {
"type": "integer",
"minimum": 1
}
},
"required": [
"chain_head",
"event_sequence"
],
"additionalProperties": false
},
"contribution_map": {
"type": "object",
"properties": {
"contribution_map_id": {
"type": "string",
"pattern": "^map_[0-9A-HJKMNP-TV-Z]{26}$"
},
"contribution_map_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
}
},
"required": [
"contribution_map_id",
"contribution_map_sha256"
],
"additionalProperties": false
},
"policy_profile": {
"type": "object",
"properties": {
"policy_profile_id": {
"type": "string",
"pattern": "^policyprofile_[0-9A-HJKMNP-TV-Z]{26}$"
},
"profile_version": {
"type": "string",
"pattern": "^\\d+\\.\\d+\\.\\d+$"
},
"profile_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
},
"retrieved_on": {
"type": "string",
"format": "date"
},
"status": {
"enum": [
"current",
"superseded",
"unavailable"
]
}
},
"required": [
"policy_profile_id",
"profile_version",
"profile_sha256",
"retrieved_on",
"status"
],
"additionalProperties": false
},
"evidentiary_status": {
"enum": [
"exact",
"degraded",
"stale",
"unverified"
]
},
"applicability_status": {
"enum": [
"current",
"stale"
]
},
"generation_status": {
"enum": [
"recorded",
"unknown"
]
},
"origin_coverage_status": {
"enum": [
"complete",
"partial",
"unknown"
]
},
"lineage_coverage_status": {
"enum": [
"complete",
"partial",
"unknown"
]
},
"identity_declaration": {
"type": "object",
"properties": {
"declared_name": {
"anyOf": [
{
"type": "string",
"minLength": 1
},
{
"type": "null"
}
]
},
"identity_status": {
"enum": [
"verified",
"self_declared",
"unknown"
]
},
"evidence_reference_ids": {
"type": "array",
"items": {
"type": "string",
"pattern": "^[a-z][a-z0-9_]*_[0-9A-HJKMNP-TV-Z]{26}$"
},
"uniqueItems": true
},
"user_approved": {
"const": true
}
},
"required": [
"declared_name",
"identity_status",
"evidence_reference_ids",
"user_approved"
],
"additionalProperties": false
},
"claim_summary": {
"type": "string",
"minLength": 1,
"maxLength": 4000
},
"ai_system_disclosures": {
"type": "array",
"items": {
"type": "object",
"properties": {
"provider_id": {
"anyOf": [
{
"type": "string",
"minLength": 1
},
{
"type": "null"
}
]
},
"model_id": {
"anyOf": [
{
"type": "string",
"minLength": 1
},
{
"type": "null"
}
]
},
"identity_status": {
"enum": [
"recorded",
"unknown"
]
},
"included_expression_segment_ids": {
"type": "array",
"items": {
"type": "string",
"pattern": "^segment_[0-9A-HJKMNP-TV-Z]{26}$"
},
"uniqueItems": true
}
},
"required": [
"provider_id",
"model_id",
"identity_status",
"included_expression_segment_ids"
],
"additionalProperties": false
}
},
"coverage": {
"type": "object",
"properties": {
"unit": {
"const": "normalized_unicode_scalar_positions"
},
"denominator": {
"type": "integer",
"minimum": 0
},
"recorded_positions": {
"type": "integer",
"minimum": 0
},
"statement": {
"type": "string",
"minLength": 1
}
},
"required": [
"unit",
"denominator",
"recorded_positions",
"statement"
],
"additionalProperties": false
},
"suggested_registration_language": {
"type": "object",
"properties": {
"author_created": {
"type": "string",
"maxLength": 1000
},
"material_excluded": {
"type": "string",
"maxLength": 1000
},
"new_material_included": {
"type": "string",
"maxLength": 1000
},
"note_to_co": {
"anyOf": [
{
"type": "string",
"maxLength": 2000
},
{
"type": "null"
}
]
},
"registration_treatment_suggestions": {
"type": "array",
"items": {
"enum": [
"claim_as_human_contribution",
"exclude_ai_generated_material",
"describe_new_human_material",
"disclose_ai_use",
"manual_review_required",
"no_suggestion"
]
},
"uniqueItems": true
},
"user_approved": {
"const": true
},
"approval_event_id": {
"type": "string",
"pattern": "^event_[0-9A-HJKMNP-TV-Z]{26}$"
},
"approved_language_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
}
},
"required": [
"author_created",
"material_excluded",
"new_material_included",
"note_to_co",
"registration_treatment_suggestions",
"user_approved",
"approval_event_id",
"approved_language_sha256"
],
"additionalProperties": false
},
"representative_transformation_operation_ids": {
"type": "array",
"items": {
"type": "string",
"pattern": "^operation_[0-9A-HJKMNP-TV-Z]{26}$"
},
"uniqueItems": true
},
"limitation_codes": {
"type": "array",
"items": {
"enum": [
"SELF_DECLARED_IDENTITY",
"UNKNOWN_IDENTITY",
"UNKNOWN_ORIGIN",
"UNKNOWN_GENERATION",
"UNKNOWN_LINEAGE",
"UNATTESTED_EXPRESSION",
"DEGRADED_COVERAGE",
"SUPERSEDED_POLICY_PROFILE",
"SANITIZED_EVIDENCE",
"COPYRIGHT_OFFICE_DETERMINES_SCOPE"
]
},
"minItems": 1,
"uniqueItems": true
},
"explanation_codes": {
"type": "array",
"items": {
"enum": [
"INTEGRITY_ONLY_VERIFICATION",
"RECORDED_ORIGIN_NOT_LEGAL_AUTHORSHIP",
"SELECTION_ARRANGEMENT_IS_OVERLAY",
"PAGE_LOCATORS_ARE_DERIVED",
"REGISTRATION_LANGUAGE_IS_SUGGESTED",
"HARP_STALE_AFTER_EDIT"
]
},
"minItems": 1,
"uniqueItems": true
},
"sanitization_profile": {
"enum": [
"full_private",
"sanitized"
]
},
"harp_sha256": {
"type": "string",
"pattern": "^sha256:[a-f0-9]{64}$"
}
},
"required": [
"schema_version",
"harp_id",
"project_id",
"application_version",
"generator_version",
"deposit",
"cpl_binding",
"contribution_map",
"policy_profile",
"evidentiary_status",
"applicability_status",
"generation_status",
"origin_coverage_status",
"lineage_coverage_status",
"identity_declaration",
"claim_summary",
"ai_system_disclosures",
"coverage",
"suggested_registration_language",
"representative_transformation_operation_ids",
"limitation_codes",
"explanation_codes",
"sanitization_profile",
"harp_sha256"
],
"additionalProperties": false,
"allOf": [
{
"if": {
"properties": {
"evidentiary_status": {
"const": "exact"
}
},
"required": [
"evidentiary_status"
]
},
"then": {
"properties": {
"applicability_status": {
"const": "current"
},
"generation_status": {
"const": "recorded"
},
"origin_coverage_status": {
"const": "complete"
},
"lineage_coverage_status": {
"const": "complete"
},
"identity_declaration": {
"type": "object",
"properties": {
"identity_status": {
"enum": [
"verified",
"self_declared"
]
}
},
"required": [
"identity_status"
]
},
"ai_system_disclosures": {
"type": "array",
"items": {
"type": "object",
"properties": {
"identity_status": {
"const": "recorded"
}
},
"required": [
"identity_status"
]
}
},
"policy_profile": {
"type": "object",
"properties": {
"status": {
"const": "current"
}
},
"required": [
"status"
]
}
}
}
}
],
"examples": [
{
"schema_version": "1.0",
"harp_id": "harp_01J00000000000000000000002",
"project_id": "project_01J00000000000000000000001",
"application_version": "0.5.2",
"generator_version": "1.0.0",
"deposit": {
"deposit_id": "deposit_01J00000000000000000000000",
"deposit_sha256": "sha256:3333333333333333333333333333333333333333333333333333333333333333",
"manuscript_revision_id": "revision_01J0000000000000000000000B",
"manuscript_revision_sha256": "sha256:2222222222222222222222222222222222222222222222222222222222222222"
},
"cpl_binding": {
"chain_head": "sha256:4444444444444444444444444444444444444444444444444444444444444444",
"event_sequence": 42
},
"contribution_map": {
"contribution_map_id": "map_01J0000000000000000000000Z",
"contribution_map_sha256": "sha256:28562ec310a4eda35436bf01ec534d0639939365af453e3c922c492ee0ebff00"
},
"policy_profile": {
"policy_profile_id": "policyprofile_01J00000000000000000000004",
"profile_version": "1.0.0",
"profile_sha256": "sha256:77bfc640282012169a92feb92b9539ca6826b798659cb3f7d07beb864e53abba",
"retrieved_on": "2026-07-19",
"status": "current"
},
"evidentiary_status": "exact",
"applicability_status": "current",
"generation_status": "recorded",
"origin_coverage_status": "complete",
"lineage_coverage_status": "complete",
"identity_declaration": {
"declared_name": "Example Author",
"identity_status": "self_declared",
"evidence_reference_ids": [
"assertion_01J0000000000000000000000V"
],
"user_approved": true
},
"claim_summary": "CPL records direct human input, accepted AI output, and subsequent recorded transformations for the exact deposit.",
"ai_system_disclosures": [
{
"provider_id": "openai",
"model_id": "gpt-example",
"identity_status": "recorded",
"included_expression_segment_ids": [
"segment_01J0000000000000000000000Y"
]
}
],
"coverage": {
"unit": "normalized_unicode_scalar_positions",
"denominator": 12,
"recorded_positions": 12,
"statement": "12 of 12 normalized Unicode scalar positions have recorded origin."
},
"suggested_registration_language": {
"author_created": "Text and revisions described in the attached provenance record",
"material_excluded": "AI-generated text identified in the attached provenance record",
"new_material_included": "Human-authored text and revisions",
"note_to_co": "This work contains disclosed AI-generated material.",
"registration_treatment_suggestions": [
"disclose_ai_use",
"manual_review_required"
],
"user_approved": true,
"approval_event_id": "event_01J00000000000000000000004",
"approved_language_sha256": "sha256:cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
},
"representative_transformation_operation_ids": [
"operation_01J0000000000000000000000X"
],
"limitation_codes": [
"SELF_DECLARED_IDENTITY",
"COPYRIGHT_OFFICE_DETERMINES_SCOPE"
],
"explanation_codes": [
"INTEGRITY_ONLY_VERIFICATION",
"RECORDED_ORIGIN_NOT_LEGAL_AUTHORSHIP",
"SELECTION_ARRANGEMENT_IS_OVERLAY",
"PAGE_LOCATORS_ARE_DERIVED",
"REGISTRATION_LANGUAGE_IS_SUGGESTED",
"HARP_STALE_AFTER_EDIT"
],
"sanitization_profile": "full_private",
"harp_sha256": "sha256:20e3ba5bb0c6c25c8d047e0e8b82a17044193e0b8ce521612d48cc85fed5feb9"
}
]
}