RDF Ingestion Source

A lightweight RDF ontology ingestion system for DataHub focused on business glossaries. This source enables ingestion of SKOS-based glossaries with term definitions, hierarchical organization, and relationships.

Overview

The RDF ingestion source provides:

• Glossary Terms: Import SKOS concepts as DataHub glossary terms
• Term Groups: Automatic creation of glossary nodes from IRI path hierarchies
• Relationships: Support for skos:broader and skos:narrower term relationships
• Standards-Based: Native support for SKOS, OWL, and RDFS vocabularies
• Modular Architecture: Pluggable entity system with auto-discovery

Quick Start

Installation

pip install acryl-datahub[rdf]

Basic Usage

Create a recipe file (rdf_glossary.yml):

source:
  type: rdf
  config:
    source: path/to/glossary.ttl
    environment: PROD

sink:
  type: datahub-rest
  config:
    server: "http://localhost:8080"
    token: "${DATAHUB_TOKEN}"

Run ingestion:

# Ingest glossary
datahub ingest -c rdf_glossary.yml

# Dry run (preview without ingesting)
datahub ingest -c rdf_glossary.yml --dry-run

With SPARQL Filter

Filter large ontologies by namespace/module:

source:
  type: rdf
  config:
    source: https://spec.edmcouncil.org/fibo/ontology/master/latest/fibo-all.ttl
    sparql_filter: |
      CONSTRUCT { ?s ?p ?o }
      WHERE {
        ?s ?p ?o .
        FILTER(STRSTARTS(STR(?s), "https://spec.edmcouncil.org/fibo/ontology/FBC/"))
      }
    environment: PROD

sink:
  type: datahub-rest
  config:
    server: "http://localhost:8080"
    token: "${DATAHUB_TOKEN}"

RDF-to-DataHub Mapping

Glossary Terms

RDF concepts are mapped to DataHub glossary terms:

• skos:Concept → GlossaryTerm
• skos:prefLabel OR rdfs:label → term name
• skos:definition OR rdfs:comment → term definition
• IRI path segments → glossary node hierarchy

Term Groups (Domains)

IRI path hierarchies are automatically converted to glossary node hierarchies:

https://example.com/finance/credit-risk
→ Glossary Node: finance
  └─ Glossary Node: credit-risk
     └─ Glossary Term: (final segment)

Note: Domains are used internally as a data structure to organize glossary terms. They are not ingested as DataHub domain entities (which are for datasets/products).

Relationships

• skos:broader → creates isRelatedTerms relationships in DataHub
• skos:narrower → creates isRelatedTerms relationships (inverse direction)

IRI-to-URN Examples

http://example.com/finance/credit-risk
→ urn:li:glossaryTerm:finance/credit-risk

fibo:FinancialInstrument
→ urn:li:glossaryTerm:fibo:FinancialInstrument

Configuration

Source Configuration

Parameter	Description	Default
source	RDF source (file, folder, URL)	required
environment	DataHub environment	PROD
format	RDF format (turtle, xml, n3, etc.)	auto-detect
dialect	RDF dialect (default, fibo, generic)	auto-detect
export_only	Export only specified types	all
skip_export	Skip specified types	none
sparql_filter	SPARQL CONSTRUCT query to filter graph	null
recursive	Recursive folder processing	true
extensions	File extensions to process	.ttl, .rdf, .owl, .n3, .nt

Export Types (CLI Options)

• glossary or glossary_terms - Glossary terms only
• relationship or relationships - Term relationships only

Note: The domain option is not available in MVP. Domains are used internally as a data structure for organizing glossary terms into hierarchies.

Example RDF File

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .

<https://example.com/finance/credit-risk>
    a skos:Concept ;
    skos:prefLabel "Credit Risk" ;
    skos:definition "The risk of loss due to a borrower's failure to repay a loan" ;
    skos:broader <https://example.com/finance/risk> .

<https://example.com/finance/risk>
    a skos:Concept ;
    skos:prefLabel "Risk" ;
    skos:definition "General category of financial risk" .

This will create:

• Glossary Node: finance
• Glossary Term: Risk (under finance node)
• Glossary Term: Credit Risk (under finance node, with relationship to Risk)

Architecture

RDF uses a modular, pluggable entity architecture:

1. Entity Extractors: Extract RDF entities from graphs
2. Entity Converters: Convert RDF AST to DataHub AST
3. MCP Builders: Generate Metadata Change Proposals (MCPs)
4. Auto-Discovery: Entity modules are automatically discovered and registered

Processing Flow

1. Load RDF files into RDF graph
2. Apply SPARQL filter (if configured) to reduce graph size
3. Extract entities (glossary terms, relationships)
4. Build domain hierarchy from IRI paths
5. Convert to DataHub AST
6. Generate MCPs for glossary nodes and terms
7. Emit to DataHub

Documentation

• RDF Specification - Complete technical specification
• Entity Plugin Contract - Guide for adding new entity types
• Documentation Index - All documentation files

Features

• ✅ Glossary Terms: Full SKOS concept support
• ✅ Term Groups: Automatic hierarchy from IRI paths
• ✅ Relationships: skos:broader/narrower support
• ✅ SPARQL Filtering: Filter RDF graphs by namespace/module before ingestion
• ✅ Multiple Formats: TTL, RDF/XML, JSON-LD, N3, N-Triples
• ✅ Multiple Sources: Files, folders, URLs
• ✅ Standards-Based: SKOS, OWL, RDFS support
• ✅ Modular: Pluggable entity architecture

MVP Scope

Current MVP includes:

• Glossary terms
• Term groups (domains) - used as data structure for hierarchy
• Term relationships

Not included in MVP:

• Datasets
• Data products
• Structured properties
• Lineage processing
• Schema fields

Requirements

• Python 3.8+
• DataHub instance
• rdflib, acryl-datahub

Getting Help

1. Start with: RDF Specification - Complete technical reference
2. Adding entities: Entity Plugin Contract - Plugin development guide
3. Examples: Review example RDF files in test fixtures
4. CLI help: Run datahub ingest --help for command options