§2.1 Storing Knowledge

What and Why

pg_ripple stores data as RDF triples — the W3C standard for representing knowledge. Every fact is a three-part statement: a subject, a predicate, and an object. This structure is deceptively simple but powerful enough to model any domain — from bibliographic records and biomedical ontologies to enterprise knowledge graphs.

Why triples instead of tables?

• Schema-free evolution: add new predicates without ALTER TABLE.
• Natural linking: every entity is an IRI — links across datasets are free.
• Standards-based: SPARQL, SHACL, OWL, and thousands of public vocabularies work out of the box.
• Provenance-ready: RDF-star lets you annotate individual facts with confidence scores, sources, and timestamps.

pg_ripple stores triples inside PostgreSQL using Vertical Partitioning (VP) — one internal table per predicate, with all values dictionary-encoded as BIGINT. You never see this machinery directly; you interact through insert_triple(), load_turtle(), and SPARQL.

How It Works

The Triple Model

Every RDF triple has the form:

<subject>  <predicate>  <object> .

• Subject: the thing you are describing (always an IRI or blank node).
• Predicate: the relationship or property (always an IRI).
• Object: the value — an IRI (another entity), a literal (string, number, date), or a blank node.

Named Graphs

Triples can be grouped into named graphs — logical partitions identified by an IRI. This is useful for:

• Tracking provenance: "these triples came from PubMed"
• Multi-tenancy: one graph per customer
• Inference output: derived triples go into a separate graph

pg_ripple uses graph ID 0 for the default graph (triples with no explicit graph). Named graphs get a positive integer ID via dictionary encoding.

Blank Nodes

Blank nodes are anonymous identifiers — they represent "something exists" without giving it a global IRI. pg_ripple encodes blank nodes with a _: prefix:

SELECT pg_ripple.insert_triple(
    '_:review1',
    '<https://schema.org/author>',
    '<https://example.org/person/alice>'
);

RDF-Star (Quoted Triples)

RDF-star lets you make statements about other statements. This is essential for provenance, confidence scores, and temporal annotations.

A quoted triple wraps << subject predicate object >> and can appear as a subject or object in another triple:

-- "The fact that Paper42 was authored by Alice has confidence 0.95"
SELECT pg_ripple.insert_triple(
    '<< <https://example.org/paper/42> <https://purl.org/dc/terms/creator> <https://example.org/person/alice> >>',
    '<https://example.org/confidence>',
    '"0.95"^^<http://www.w3.org/2001/XMLSchema#decimal>'
);

Dictionary Encoding

Every IRI, blank node, and literal is mapped to a BIGINT (i64) via XXH3-128 hashing before storage. VP tables contain only integers — this makes joins fast and storage compact. You never need to think about encoding; pg_ripple handles it transparently.

Worked Examples

The examples in this chapter use a bibliographic dataset: papers, authors, institutions, journals, and citations.

Setting Up Prefixes

Register namespace prefixes so SPARQL queries are readable:

SELECT pg_ripple.register_prefix('ex',    'https://example.org/');
SELECT pg_ripple.register_prefix('dct',   'http://purl.org/dc/terms/');
SELECT pg_ripple.register_prefix('foaf',  'http://xmlns.com/foaf/0.1/');
SELECT pg_ripple.register_prefix('bibo',  'http://purl.org/ontology/bibo/');
SELECT pg_ripple.register_prefix('schema','https://schema.org/');
SELECT pg_ripple.register_prefix('xsd',   'http://www.w3.org/2001/XMLSchema#');

Inserting Individual Triples

-- Create a paper
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>',
    '<http://purl.org/ontology/bibo/AcademicArticle>'
);

-- Add a title
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/title>',
    '"Knowledge Graphs in Practice"'
);

-- Add an author
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/creator>',
    '<https://example.org/person/alice>'
);

-- Author metadata
SELECT pg_ripple.insert_triple(
    '<https://example.org/person/alice>',
    '<http://xmlns.com/foaf/0.1/name>',
    '"Alice Johnson"'
);

SELECT pg_ripple.insert_triple(
    '<https://example.org/person/alice>',
    '<https://schema.org/affiliation>',
    '<https://example.org/institution/mit>'
);

-- Institution metadata
SELECT pg_ripple.insert_triple(
    '<https://example.org/institution/mit>',
    '<http://xmlns.com/foaf/0.1/name>',
    '"Massachusetts Institute of Technology"'
);

Loading a Full Dataset with Turtle

For bulk data, Turtle format is more natural:

SELECT pg_ripple.load_turtle('
@prefix ex:     <https://example.org/> .
@prefix dct:    <http://purl.org/dc/terms/> .
@prefix foaf:   <http://xmlns.com/foaf/0.1/> .
@prefix bibo:   <http://purl.org/ontology/bibo/> .
@prefix schema: <https://schema.org/> .
@prefix xsd:    <http://www.w3.org/2001/XMLSchema#> .

ex:paper/42 a bibo:AcademicArticle ;
    dct:title "Knowledge Graphs in Practice" ;
    dct:creator ex:person/alice, ex:person/bob ;
    dct:date "2024-03-15"^^xsd:date ;
    bibo:citedBy ex:paper/99 ;
    schema:keywords "knowledge graph", "RDF", "SPARQL" .

ex:paper/99 a bibo:AcademicArticle ;
    dct:title "Graph Neural Networks for Entity Resolution" ;
    dct:creator ex:person/carol ;
    bibo:cites ex:paper/42 .

ex:person/alice foaf:name "Alice Johnson" ;
    schema:affiliation ex:institution/mit .

ex:person/bob foaf:name "Bob Smith" ;
    schema:affiliation ex:institution/stanford .

ex:person/carol foaf:name "Carol Williams" ;
    schema:affiliation ex:institution/mit .

ex:institution/mit foaf:name "Massachusetts Institute of Technology" .
ex:institution/stanford foaf:name "Stanford University" .
');

Using Named Graphs

Store triples from different sources in separate graphs:

-- Create named graphs for different data sources
SELECT pg_ripple.create_graph('https://example.org/graph/pubmed');
SELECT pg_ripple.create_graph('https://example.org/graph/arxiv');

-- Load PubMed data into its graph
SELECT pg_ripple.load_turtle_into_graph('
@prefix ex:   <https://example.org/> .
@prefix dct:  <http://purl.org/dc/terms/> .
@prefix bibo: <http://purl.org/ontology/bibo/> .

ex:paper/100 a bibo:AcademicArticle ;
    dct:title "Drug Interaction Networks" ;
    dct:creator ex:person/dave .
', 'https://example.org/graph/pubmed');

-- Load arXiv data into its graph
SELECT pg_ripple.load_turtle_into_graph('
@prefix ex:   <https://example.org/> .
@prefix dct:  <http://purl.org/dc/terms/> .
@prefix bibo: <http://purl.org/ontology/bibo/> .

ex:paper/200 a bibo:AcademicArticle ;
    dct:title "Transformer Architectures for NLP" ;
    dct:creator ex:person/eve .
', 'https://example.org/graph/arxiv');

-- List all named graphs
SELECT * FROM pg_ripple.list_graphs();

RDF-Star for Provenance and Confidence

Annotate citations with provenance metadata:

-- Record that Paper 42 cites Paper 99 (the base fact)
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/ontology/bibo/cites>',
    '<https://example.org/paper/99>'
);

-- Annotate this citation with a confidence score
SELECT pg_ripple.insert_triple(
    '<< <https://example.org/paper/42> <http://purl.org/ontology/bibo/cites> <https://example.org/paper/99> >>',
    '<https://example.org/confidence>',
    '"0.92"^^<http://www.w3.org/2001/XMLSchema#decimal>'
);

-- Record who asserted this citation
SELECT pg_ripple.insert_triple(
    '<< <https://example.org/paper/42> <http://purl.org/ontology/bibo/cites> <https://example.org/paper/99> >>',
    '<http://purl.org/dc/terms/source>',
    '<https://example.org/system/citation-extractor>'
);

Translating a Relational Schema to RDF

Suppose you have a relational database with tables papers, authors, and affiliations:

The mapping pattern:

1. Each row becomes a subject IRI: <https://example.org/paper/{id}>
2. Each column becomes a predicate: use a standard vocabulary (Dublin Core, Schema.org, FOAF)
3. Foreign keys become object IRIs: authors.institution_id = 10 → <https://example.org/institution/10>
4. Scalar values become literals: papers.title → "Knowledge Graphs in Practice"

-- Row from papers table → triples
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>',
    '<http://purl.org/ontology/bibo/AcademicArticle>'
);
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/title>',
    '"Knowledge Graphs in Practice"'
);
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/date>',
    '"2024"^^<http://www.w3.org/2001/XMLSchema#gYear>'
);

-- Foreign key → IRI link
SELECT pg_ripple.insert_triple(
    '<https://example.org/person/1>',
    '<https://schema.org/affiliation>',
    '<https://example.org/institution/10>'
);

Common Patterns

Pattern: Type Hierarchies

Use rdf:type and rdfs:subClassOf to create type hierarchies:

SELECT pg_ripple.load_turtle('
@prefix ex:   <https://example.org/> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix bibo: <http://purl.org/ontology/bibo/> .

bibo:AcademicArticle rdfs:subClassOf bibo:Article .
bibo:Article rdfs:subClassOf bibo:Document .

ex:paper/42 a bibo:AcademicArticle .
');

With RDFS inference enabled (see §2.5), pg_ripple can automatically derive that ex:paper/42 is also a bibo:Article and a bibo:Document.

Pattern: Multi-Valued Properties

Unlike relational columns, RDF predicates are naturally multi-valued:

-- A paper can have multiple authors — just insert multiple triples
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/creator>',
    '<https://example.org/person/alice>'
);
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/creator>',
    '<https://example.org/person/bob>'
);

Pattern: Typed and Language-Tagged Literals

-- Typed literal (date)
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/date>',
    '"2024-03-15"^^<http://www.w3.org/2001/XMLSchema#date>'
);

-- Language-tagged string
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/title>',
    '"Knowledge Graphs in Practice"@en'
);
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/title>',
    '"Wissensgraphen in der Praxis"@de'
);

Pattern: Reification with RDF-Star vs Named Graphs

Two approaches for tracking who said what:

RDF-star — annotate individual triples:

SELECT pg_ripple.insert_triple(
    '<< <https://example.org/paper/42> <http://purl.org/dc/terms/creator> <https://example.org/person/alice> >>',
    '<http://purl.org/dc/terms/source>',
    '<https://example.org/dataset/pubmed>'
);

Named graphs — group triples by source:

SELECT pg_ripple.load_turtle_into_graph('
@prefix ex:  <https://example.org/> .
@prefix dct: <http://purl.org/dc/terms/> .
ex:paper/42 dct:creator ex:person/alice .
', 'https://example.org/dataset/pubmed');

Use RDF-star when different triples about the same entity have different provenance. Use named graphs when entire batches share the same source.

Performance and Trade-offs

• Dictionary cache: frequently used IRIs (predicates, common types) stay in the shared-memory LRU cache. Check hit rates with SELECT pg_ripple.cache_stats().
• VP table promotion: predicates with fewer than 1,000 triples share the vp_rare consolidation table. Once a predicate crosses the threshold, it gets its own dedicated VP table with dual B-tree indexes.
• Named graph overhead: the g column adds 8 bytes per triple. If you do not need named graphs, using the default graph (the default) avoids the cost of graph-ID lookups.

SELECT pg_ripple.vacuum();

</div>
</div>

-- WRONG: will be treated as a plain literal
SELECT pg_ripple.insert_triple(
    'https://example.org/paper/42',
    'http://purl.org/dc/terms/title',
    '"Hello"'
);

-- CORRECT: angle brackets around IRIs
SELECT pg_ripple.insert_triple(
    '<https://example.org/paper/42>',
    '<http://purl.org/dc/terms/title>',
    '"Hello"'
);

-- Plain string
'"Hello"'

-- Typed integer
'"42"^^<http://www.w3.org/2001/XMLSchema#integer>'

-- Language-tagged string
'"Bonjour"@fr'

-- All triples about Paper 42
SELECT * FROM pg_ripple.find_triples(
    '<https://example.org/paper/42>', NULL, NULL
);

-- All triples with the dct:creator predicate
SELECT * FROM pg_ripple.find_triples(
    NULL, '<http://purl.org/dc/terms/creator>', NULL
);

-- Total triple count
SELECT pg_ripple.triple_count();