Begrippenkader Schema YAML

Begrippenkader Schema YAML

Guide for editing the begrippenkader (sources/begrippenkader_dpia.yaml) that conforms to schemas/begrippenkader.v1.schema.json.

There is a second begrippenkader for the IAMA assessment, sources/begrippenkader_iama.yaml, used to enrich sources/iama.yaml. It is generated, not hand-edited — see IAMA begrippenkader below.

Top-level Structure

schema_version: 1.0.0
name: "Begrippenkader Model DPIA Rijksdienst en Pre-scan DPIA"
description: "..."
urn: "urn:nl:begrippenkaderdpia_pre-scandpia:3.0:begrippenkader:1.0.0"
language: nl
version_date: "2025-06-01"
owners:
  - organization: "Ministerie van Binnenlandse Zaken"
    name: "CIO Rijk"
    email: "[email protected]"
    role: "Privacy Adviseurs Rijk"
definitions: [...]                # Array of definition objects

Definition Structure

- id: verwerker                   # Required: lowercase letters, digits, underscores only
  term: Verwerker                 # Required: display name with proper capitalization
  category: "DPIA - 06. Betrokken partijen"  # Required: category string
  definition: >-                  # Required: definition text
    Een verwerker is een natuurlijke persoon of rechtspersoon die
    ten behoeve van de [verwerkingsverantwoordelijke] [persoonsgegevens] verwerkt.
  metadata:                       # Optional metadata
    kennisbronnen:
      - "artikel 4, lid 8, AVG"
    toelichting: "Uitgebreide uitleg..."
    redactionele_opmerking: "Bron van deze definitie..."
    voorbeelden:
      - "Een cloudprovider die data opslaat"
    alternatieve_spellingen:
      - "processor"
    meervoudsvormen:
      - "verwerkers"

ID Convention

• Pattern: ^[a-z0-9_]+$ — only lowercase letters, digits, and underscores
• Use underscores to separate words: recht_van_inzage, lijst_edpb_geautomatiseerde_besluitvorming
• Keep IDs descriptive but concise

Categories

Categories follow these patterns:

• DPIA sections: "DPIA - <number>. <section name>" (e.g. "DPIA - 06. Betrokken partijen")
• Pre-scan sections: "Pre-scan DPIA - <letter>. <section name>" (e.g. "Pre-scan DPIA - E. Lijst EDPB")

Cross-references in Definitions

Terms enclosed in [vierkante haken] (square brackets) within definition text reference other definitions in the begrippenkader:

definition: >-
  Een verwerker is een natuurlijke persoon die ten behoeve van de
  [verwerkingsverantwoordelijke] [persoonsgegevens] verwerkt.

Here [verwerkingsverantwoordelijke] and [persoonsgegevens] link to their respective definitions. When adding new definitions, use square brackets to reference existing terms.

Metadata Fields

Field	Type	Description
kennisbronnen	string[]	Legal/document sources (e.g. "artikel 15, lid 1, AVG")
toelichting	string	Extended explanation
redactionele_opmerking	string	Editorial note about the definition's origin
voorbeelden	string[]	Concrete examples
alternatieve_spellingen	string[]	Alternative spellings (e.g. English equivalent)
meervoudsvormen	string[]	Plural forms of the term

Definition Enrichment Pipeline

The script/definition_enricher.py script processes definitions for use in the frontend:

1. Creates a term map from all definitions (including plural forms and alternative spellings)
2. Injects <span class="aiv-definition"> HTML tags around matching terms in assessment YAML
3. Matching priority: hoofdtermen → meervoudsvormen → alternatieve spellingen → alternatieve termen
4. All matching is case-insensitive; original capitalization is preserved
5. Existing definition spans are not re-processed (no nesting)

After editing the begrippenkader, test enrichment:

python script/definition_enricher.py \
  --source sources/dpia.yaml \
  --definitions sources/begrippenkader_dpia.yaml \
  --output /tmp/test_enriched.json

Schema Validation

python script/schema_validator.py \
  --schema schemas/begrippenkader.v1.schema.json \
  --source sources/begrippenkader_dpia.yaml

Note: The schema expects glossary[] but the actual YAML uses definitions[]. The schema may need updating to match. Always check the actual file structure alongside the schema.

IAMA begrippenkader

The IAMA assessment uses a separate begrippenkader, sources/begrippenkader_iama.yaml, which enriches sources/iama.yaml (urn urn:nl:iama, version 2.0).

• Generated, not hand-edited. It is produced from the Algoritmekader begrippenlijst via script/convert_definitions_from_algoritmekader.py. Edit the source begrippenlijst / regenerate rather than editing the YAML by hand.
• Simpler structure than the DPIA begrippenkader: a flat definitions[] list where each entry has only term and definition — no id, category, metadata, top-level urn, or owners.

definitions:
  - term: aanbieder
    definition: >-
      Een natuurlijke of rechtspersoon, overheidsinstantie, agentschap of ander
      orgaan die/dat een AI-systeem ... ontwikkelt of laat ontwikkelen ...

Regenerate with:

python script/convert_definitions_from_algoritmekader.py --output sources/begrippenkader_iama.yaml

When generating sources/generated/IAMA.json, enrichment runs with the --definitions-once-per-page flag, so each term tooltip is injected at most once per "deel" instead of at every occurrence.

Statistics

The DPIA/Pre-scan begrippenkader currently contains 456 definitions; the IAMA begrippenkader (begrippenkader_iama.yaml) contains 95 definitions from the Algoritmekader.