Meta Museum Docs (Live Markdown Source)

Based on my comprehensive research, here’s the expanded section on Vocabularies and Authority Files:

Vocabularies and Authority Files

Getty Vocabularies (as LOD)

The Getty Vocabularies are comprehensive, multilingual controlled vocabularies for cultural heritage, available as Linked Open Data with extensive technical infrastructure:

Art & Architecture Thesaurus (AAT)

• Content: 57,390+ hierarchical concepts covering generic terms for art, architecture, decorative arts, materials, techniques, styles, and conservation

• Scope: Work types, roles, materials, styles, cultures, techniques (e.g., amphora, oil paint, sintering)

• Multiple languages with ongoing expansion for multilingual and multicultural coverage

• URI pattern: http://vocab.getty.edu/aat/[ID]

• Example: http://vocab.getty.edu/aat/300198841 (rhyta)

Union List of Artist Names (ULAN)

• Content: 367,590+ records with names, biographical information, and relationships

• Covers: Artists, architects, firms, studios, repositories, patrons, both named and anonymous entities

• Includes variant names across languages and time periods

• URI pattern: http://vocab.getty.edu/ulan/[ID]

• Example: http://vocab.getty.edu/ulan/500115493 (Albrecht Dürer)

• Links to biographical information at: http://vocab.getty.edu/ulan/[ID]-agent

Getty Thesaurus of Geographic Names (TGN)

• Hierarchical geographic vocabulary

• Current and historical place names with coordinates

• Multiple language variants

• URI pattern: http://vocab.getty.edu/tgn/[ID]

• Example: http://vocab.getty.edu/tgn/7011179 (Siena)

• Geographic info at: http://vocab.getty.edu/tgn/[ID]-place

Cultural Objects Name Authority (CONA)

• Focus: Architecture, multiples, works depicted in other works

• Includes: Titles, artist attributions, creation dates, locations (extant or destroyed)

• Example works: Hagia Sophia, The Lacemaker, Great Wave off Kanagawa

• URI pattern: http://vocab.getty.edu/page/cona/[ID]

• Available through APIs (not yet as full LOD)

Getty Iconography Authority (IA)

• Proper names for iconographical narratives, religious/fictional characters, historical events

• Literary works and performing arts

• Special focus on non-Western subjects

• Coreferences to Iconclass for Western iconography

• Examples: Shiva, Avalokiteshvara, French Revolution, Xibalba

• Available through APIs (included with CONA Web Services)

Access Methods:

• SPARQL endpoint: https://data.getty.edu/vocab/sparql

• SPARQL UI: https://data.getty.edu/vocab/sparql-ui

• Activity Stream: https://data.getty.edu/vocab/activity-stream

• Full downloads: N-Triples format (AAT, TGN, ULAN), refreshed monthly

• Individual records: JSON, RDF/XML, N3/Turtle, N-Triples

• OpenRefine reconciliation service available

• NEW: Linked.Art semantic representation as default for JSON/JSON-LD

Important Notes:

• License: Open Data Commons Attribution License (ODC-By) 1.0

• XML and Relational Tables formats being retired end of 2025

• LOD is the recommended sustainable format

• Over 90 sample SPARQL queries available in documentation

• Documentation: https://www.getty.edu/research/tools/vocabularies/lod/

VIAF (Virtual International Authority File)

VIAF is a collaborative international authority file that virtually combines name authority files from national libraries and other institutions worldwide.

Overview:

• Managed by OCLC in cooperation with 37+ agencies from 29+ countries

• Combines: Library of Congress, Deutsche Nationalbibliothek, Bibliothèque nationale de France, and many others

• 8+ million authority records for persons, corporate bodies, and works

• Serves as an identity hub linking to multiple other authority sources

Content:

• Personal names with variant forms across languages and scripts

• Corporate bodies

• Conference names

• Uniform titles

• Biographical information from contributing sources

URI Structure:

• Canonical URI: https://viaf.org/viaf/[numerical_ID]

• Example: https://viaf.org/viaf/15873 (Pablo Picasso)

• Each VIAF record asserts owl:sameAs relationships to corresponding entities in LC, Getty, Wikidata, DBpedia, etc.

Access Methods:

• VIAF API: Search and retrieve authority records

• https://www.oclc.org/developer/api/oclc-apis/viaf.en.html

• Supports keyword search, control number lookup

• Returns cluster records and source records

• Content negotiation: XHTML/RDFa, RDF/XML, N-Triples, JSON

• SRU search interface

• Complete data dumps: http://viaf.org/viaf/data/

• Available in RDF, MARC-21, plain text

• Files last updated August 2024 (currently not being updated during infrastructure improvements)

• OpenRefine reconciliation endpoints available from third parties:

• http://refine.codefork.com/reconcile/viaf

• http://iphylo.org/~rpage/phyloinformatics/services/reconciliation_viaf.php

Important Notes:

• License: Open Data Commons Attribution License (ODC-By) 1.0

• No public SPARQL endpoint (though data available for local processing)

• HDT (Header Dictionary Triples) compressed versions available for efficient querying

• Experimental Linked Data Fragments servers available

• VIAF incorporated in ULAN and being integrated into other Getty vocabularies

GeoNames

GeoNames is a comprehensive geographical database with worldwide coverage, available as Linked Open Data.

Overview:

• 11+ million geographical names

• 9+ million unique features

• 2.8+ million populated places

• Multilingual coverage

Content:

• Place names with coordinates (WGS84)

• Administrative divisions

• Postal codes

• Population data

• Feature classifications (mountains, rivers, cities, etc.)

• Alternate names in multiple languages

• Hierarchical relationships between places

Ontology:

• GeoNames Ontology: https://www.geonames.org/ontology/ontology_v3.3.rdf

• Based on W3C RDF and OWL standards

• Compatible with WGS84 positioning vocabulary

URI Structure:

• Uses 303 redirection to distinguish concept from document

• Concept URI: https://sws.geonames.org/[ID]/

• RDF document: https://sws.geonames.org/[ID]/about.rdf

• Example: https://sws.geonames.org/3020251/ (Embrun, France)

Access Methods:

• RDF/XML web service for individual features

• REST API for searches and queries

• SPARQL queries via third-party endpoints:

• GeoSPARQL.org: https://www.geosparql.org/

• FactForge: http://factforge.net/repositories/ff-news

• Full data dumps available for download

• OpenRefine reconciliation service

Formats:

• RDF/XML

• Turtle

• JSON

• GeoJSON for spatial queries

Integration:

• Widely linked in LOD cloud

• Used by DBpedia, Linked Geodata, and other major LOD datasets

• Compatible with GeoSPARQL for spatial queries

Library of Congress Linked Data Service (id.loc.gov)

The Library of Congress provides comprehensive authority data and vocabularies as Linked Open Data.

Available Datasets:

LC Name Authority File (LCNAF)

• 12+ million name authority records

• Persons, families, corporate bodies

• Conference names

• URI pattern: http://id.loc.gov/authorities/names/[ID]

Library of Congress Subject Headings (LCSH)

• Comprehensive subject authority file

• Hierarchical relationships (broader, narrower, related terms)

• Cross-linked with RAMEAU (French equivalent)

• URI pattern: http://id.loc.gov/authorities/subjects/[ID]

LC Classification (LCC)

• Complete classification scheme

• URI pattern: http://id.loc.gov/authorities/classification/[ID]

Other Vocabularies:

• LC Genre/Form Terms (LCGFT)

• LC Demographic Group Terms (LCDGT)

• LC Medium of Performance Thesaurus (LCMPT)

• Thesaurus for Graphic Materials (TGM)

• MARC code lists and relators

• BIBFRAME vocabularies

Access Methods:

• Search interface: https://id.loc.gov/

• Suggest APIs (Suggest and Suggest2)

• Search API with advanced query parameters:

• memberOf, scheme, contentSource, rdftype filters

• Token search for LCCNs

• Date filters (cdate, mdate)

• Content negotiation for individual records

• No SPARQL endpoint currently available

Formats:

• MADS/RDF and SKOS (where appropriate)

• XHTML/RDFa

• RDF/XML

• N-Triples

• JSON-LD

• Bulk downloads available in all formats

Important Features:

• Links to external sources (Wikidata, GND, Getty, VIAF)

• Updated regularly

• API documentation: https://id.loc.gov/techcenter/

• License: Public domain

Iconclass

Iconclass is the most widely accepted international classification system for iconographic research and subject indexing of visual arts.

Overview:

• 28,000+ hierarchically ordered definitions

• 14,000+ keywords in alphabetical index

• Multilingual: English, German, French, Italian, Finnish, Norwegian, with experimental Chinese and Dutch

• Developed by Henri van de Waal (1940s-1985)

• Maintained by Henri van de Waal Foundation and Iconclass Consortium

Structure:

• 10 main divisions with alphanumeric notation codes

• Highly detailed hierarchies for Biblical and mythological subjects

• Expandable system with “keys” creating 1+ million possible notations

• WITH-NAMES placeholders (e.g., 11H(…) for male saints)

• Example notation: 71H7131 = “Bathsheba (alone) with David’s letter”

Content Coverage:

• Religious and mythological subjects

• Historical events and narratives

• Secular subjects, objects, and concepts

• Abstract ideas

• Actions and events depicted in art

• Pre-iconographical through iconological interpretation levels

Access as LOD:

• Base URI: http://iconclass.org/[notation]

• JSON format: http://iconclass.org/[notation].json

• RDF/SKOS format: http://iconclass.org/[notation].rdf

• Example: http://iconclass.org/52D1 or http://iconclass.org/52D1.rdf

Data Dumps:

• Complete SKOS JSON-LD NDJSON dump: http://iconclass.org/data/iconclass_20200710_skos_jsonld.ndjson.gz

• ~140MB compressed, 2GB+ uncompressed

• GitHub repository: https://github.com/iconclass/data

• Python library available

APIs:

• Batch notation lookup: http://iconclass.org/json/?notation=25FF&notation=51A

• Natural language search: http://iconclass.org/rkd/0/?q=lion&q_s=1&fmt=json

• Proper URL encoding required for special characters

Integration:

• Links to AAT concepts for overlapping terminology

• Integrated with Getty Iconography Authority

• Used by major museums worldwide (Gemäldegalerie Berlin, RKD, etc.)

• BARTOC entry: https://bartoc.org/en/node/459

Important Notes:

• Free access guaranteed by Iconclass Foundation

• Consortium membership available from July 2024

• Extensive bibliography (7 volumes, not yet online)

• Browser interface: https://iconclass.org/

GND (Gemeinsame Normdatei / Integrated Authority File)

GND is the integrated authority file for German-speaking countries, increasingly used internationally.

Overview:

• 9+ million authority records

• Maintained by German National Library (DNB) and GND Cooperative

• Merges former separate files: PND (personal names), GKD (corporate bodies), SWD (subject headings), EST (uniform titles)

• Used by libraries, archives, museums, and research projects

Entity Types:

• Persons (differentiated persons)

• Corporate bodies and organizations

• Conferences and events

• Geographic names and places

• Subject headings

• Works (uniform titles)

Access Methods:

• SPARQL endpoint (BETA): https://sparql.dnb.de/

• Lobid-GND service: http://lobid.org/gnd/

• Search interface

• JSON-LD API

• OpenRefine reconciliation: http://lobid.org/gnd/reconcile/

• Direct data access: https://data.dnb.de/opendata/

• OAI-PMH updates service

URI Structure:

• Authority URI: https://d-nb.info/gnd/[GND-ID]

• Example: https://d-nb.info/gnd/118540238 (Johann Wolfgang von Goethe)

Formats Available:

• RDF/XML (complete downloads by entity type)

• MARC 21 Authority Format

• MARC21-XML

• JSON-LD

• Turtle

• EntityFacts (JSON-LD subset)

GND Ontology:

• Specification: https://d-nb.info/standards/elementset/gnd

• RDF vocabulary for Semantic Web use

• CC0 1.0 Universal license

• Scheduled releases: January, May, September

• Mailing list for updates: http://lists.dnb.de/mailman/listinfo/gnd-ontology

Data Downloads:

• Full dumps by category (geografikum, koerperschaft, kongress, person, sachbegriff, werk)

• Available as .rdf.gz files

• Change feed for incremental updates

• Last update tracking

Important Features:

• Links to VIAF, Wikidata, and other international authority files

• Multilingual with focus on German

• Expanding beyond library use to broader cultural sector

• Part of NFDI4Culture initiative

• License: CC0 1.0 Universal (for ontology); various for data

Integration:

• Incorporated into VIAF

• Cross-linked with LC authorities

• Links to Getty vocabularies

• Wikidata integration

All of these vocabularies and authority files use standard Semantic Web technologies and are increasingly interconnected through owl:sameAs and other linking properties, forming a rich ecosystem of cultural heritage knowledge representation.​​​​​​​​​​​​​​​​

┌──────────────────────────┐

│ Frontend (Next.js 15) │

│ - Artwork/Entity pages │

│ - Auto-suggest fields │

│ - LOD enriched narrative│

└─────────────▲────────────┘

│ GraphQL/REST

┌─────────────┴────────────┐

│ API Gateway │

│ (.NET Aspire or Node) │

│ - LOD aggregation │

│ - Caching (Redis) │

│ - AI enrichment │

└─────────────▲────────────┘

│ SPARQL / APIs

┌─────────────┴─────────────────────┐

│ LOD Connectors / Microservices │

│ - Getty SPARQL │

│ - VIAF API │

│ - GeoNames API │

│ - LC suggest API │

│ - Iconclass API │

│ - LDAP / Wikidata │

└─────────────▲─────────────────────┘

│ URIs resolve

┌─────────────┴────────────────────┐

│ Triple Store (optional) │

│ Blazegraph / GraphDB / Jena │

│ - Stores enriched data │

│ - Fast semantic queries │

└─────────────▲────────────────────┘

│

┌─────────────┴────────────────────┐

│ AI Agents Layer │

│ - RAG narrative generator │

│ - Link discovery │

│ - Entity classification │

│ - Iconography suggestions │

└───────────────────────────────────┘

🌐 How to Use Getty, VIAF, LC, GeoNames, GND, Iconclass, Linked.Art, and AI Agents Together in a Modern Web App

Full technical architecture, full data flow, and implementation strategy.

No tables. Fully detailed. Deep technical content.

1. Conceptual Model: URIs as the Backbone

A Linked Open Data web app does not merge datasets; instead it models your own domain objects (artworks, people, places, events, specimens, archives, books, etc.) and attaches URIs from established authority vocabularies.

Each vocabulary governs a specific knowledge domain.

Your local app stores your own JSON-LD records, but the “meaning” is always delegated to external URIs.

An artwork references:

– AAT URIs for style, technique, material, object type

– ULAN/VIAF/LCNAF/GND for people

– TGN or GeoNames for places

– Iconclass or Getty IA for iconographic concepts

– CONA or other work authorities for related works

– LCSH or other subject vocabularies for subject descriptions

Your application becomes an orchestrator of distributed linked meaning.

1. Base Format: JSON-LD using Linked.Art

The data format that unifies everything is JSON-LD using the Linked.Art profile.

Linked.Art is built on CIDOC-CRM (semantic model) but expressed in a developer-friendly JSON-LD frame with stable classes like HumanMadeObject, Activity, Person, Place, Material, and Type.

A typical record includes URIs rather than string literals for anything with conceptual meaning.

The app only stores its own URI plus URIs pointing outward.

Everything else—labels, descriptions, biographical data, multilingual names—is dereferenced from the authority services.

1. UI/UX and Data Fetching: URI Resolution Layer

The core of integrating multiple vocabularies is a URI resolver microservice within your web app architecture.

This service receives any external URI and returns normalized JSON.

1. Front-End Architecture: Next.js 15 with Server Actions

The front-end (using React Server Components + Server Actions) fetches enriched data only when needed. You never fetch RDF directly in the browser.

The flow:

1. Artwork page loads server-side.

1. Artwork JSON-LD is retrieved from your database or object store.

1. Server calls /lod/resolve for any URIs attached to the artwork:

– ULAN → person info

– AAT → terms (material, technique, etc.)

– TGN → place + coordinates

– Iconclass → iconography

– VIAF → crosslinks and variant names

– LC → subjects, work categories

1. Server renders the page from enriched metadata.

1. The browser receives a fully hydrated React component tree.

This results in:

– blazing fast load times

– SEO-compatible semantic HTML

– no CORS or RDF parsing in the client

1. SPARQL + API Integration Layer

Your back end uses the following integration strategies:

Getty AAT / ULAN / TGN:

– Query using the Getty SPARQL endpoint

– Content negotiation for Turtle/JSON-LD

– TGN geometry parsed into WGS84 coordinates

GND:

– SPARQL endpoint (sparql.dnb.de) for typed relationships

– GND-ontology URIs mapped to RDF types

– Lobid JSON-LD API for human-readable data

VIAF:

– XML or JSON cluster retrieval

– Matching algorithms for variant names

– Internal caching due to VIAF latency

Iconclass:

– SKOS-based RDF through .rdf files

– JSON API for multi-notation lookup

– Use Iconclass “expanded_notation” for better faceting

LC Authorities:

– id.loc.gov Suggest API for autosuggest

– Content-negotiation to JSON-LD for individual entities

– MARC relators mapped to CRM properties

GeoNames:

– RDF endpoint for authoritative concepts

– REST API for population and feature code

– GeoJSON conversion for mapping UI

All responses are normalized through internal vocabulary adapters (similar to how Wikidata’s “Frogmouth” model normalizes varied sources).

1. AI Agent Integration (The Critical Layer)

AI agents operate across three modes:

Mode 1: Entity Extraction from Text

Extract person names, materials, styles, locations, iconography keywords.

Mode 2: Entity Reconciliation

For each extracted name or concept:

– Query ULAN and VIAF

– Score matches by lexical similarity, date overlap, role match

– Disambiguate “John Smith” using birth-death dates

– Link materials/techniques to AAT using cosine similarity on embeddings

Mode 3: Narrative Construction

Once entities are reconciled, the agent can produce:

– Biographical context

– Iconographic interpretation

– Material analysis

– Geographic mapping

– Historical timeline

The agent uses:

– your triple store

– your resolver cache

– Wikidata and Getty cross-links

– a local vector index of enriched metadata

– a narrative schema for consistent output

This creates dynamic interpretation layers over static LOD.

1. Optional Triple Store Layer for Semantic Reasoning

A triple store is not mandatory but gives you:

1. SPARQL for multi-entity analytics

1. Semantic joins across vocabularies

1. Efficient reasoning for:

– inference of broader/narrower AAT types

– geographic containment (TGN/GeoNames)

– person roles and relationships (ULAN/GND)

1. Long-term caching of authority data

1. Integration with a RAG pipeline

When included, the triple store populates itself by:

– ingesting remote RDF from vocabularies

– ingesting your local JSON-LD

– running inference rules (Jena or GraphDB reasoners)

This becomes the “semantic memory” for your AI agents.

1. Putting Everything Together on One Website

A fully integrated LOD web app behaves like this:

1. User visits /artwork/123.

1. Next.js server returns your base JSON-LD record.

1. Server identifies external URIs and calls your resolver microservice.

1. Resolver fetches Getty, VIAF, LC, GND, Iconclass, or GeoNames data.

1. Data is normalized into a clean, consistent JSON structure.

1. Your page includes:

– artist identity with cross-links

– biography with verified dates

– material and technique definitions from AAT

– iconography interpretation from Iconclass

– map visualization using GeoNames or TGN coordinates

– subject headings from LC

– hierarchical browsing (AAT broader/narrower terms)

1. AI agent injects:

– story of the work

– related works

– temporal and biographical contextualization

– inferred relationships not explicitly stated

Your website becomes a narrative knowledge graph powered by real LOD authorities.

1. If You Want, I Can Now Generate

– the complete code architecture

– the microservices

– the Next.js data fetching layer

– the .NET Aspire integration layer

– the resolver service

– the triple store ingestion scripts

– the AI agents for narrative enrichment

– or a fully deployable GitHub template

Just tell me:

“Generate the full LOD web app codebase.”