ConceptDBISAI-written · not yet proofread

RDF Triple

Also RDF Triples

An RDF triple `subject predicate object .` is the single unit of data in RDF; a set of triples forms a knowledge graph, a directed graph with one node per entity and one predicate-labelled edge per triple.

Definition

The Resource Description Framework (RDF) models all data as a set of triples

subject predicate object .

where the subject is an entity, the predicate is a binary relation, and the object is an entity or a literal (a value). Each triple states one fact: the subject stands in relation predicate to the object.

A set of triples is a knowledge graph (KG), read as a directed labelled graph:

nodes: each distinct entity (and literal) is one node,
edges: each triple is a directed edge from the subject node to the object node, labelled with the predicate.

Properties#

Identifiers. Entities and predicates are named by IRIs (International Resource Identifiers): globally unique, URL-like, but Unicode characters (ä, ö, ü, …) are allowed. IRIs are often ID-like (wd:Q37459 = Nicole Kidman); human-readable names are attached by separate triples, e.g. with rdfs:label.
Literals. An object can be a literal instead of an IRI: a string with optional language tag ("Nicole Kidman"@en) or a typed value ("1967-06-20"^^xsd:date).
Prefixes. Long IRIs are abbreviated by prefixes, e.g. wd: = <http://www.wikidata.org/entity/>, so wd:Q18094 = <http://www.wikidata.org/entity/Q18094> (see Wikidata Prefix System).
Schema-free. There is no fixed schema: any entity may have any predicate, and triples can share subjects, predicates and objects.
Only binary relations. A triple relates exactly two things. Facts with more components ( $n$ -ary relations, $n > 2$ , e.g. award + year + film) need Reification.
Interoperability. Two datasets are combined by adding a few triples that relate their IDs, e.g. wd:Q183 wdtn:P402 osmrel:51477 . (Germany in Wikidata ↔ Germany in OpenStreetMap). This enables federated queries over several KGs, which is hard for relational databases with different schemas and identifiers.
Storage in a relational DB. A KG fits into one table triples(subject, predicate, object), or into one two-column table per predicate; this is the basis of SPARQL to SQL Translation.
Large KGs: Wikidata (21 B triples, 62 k predicates), UniProt (240 B), PubChem (26 B), OpenStreetMap (202 B), DBLP (1.5 B).

Example#

<Nicole_Kidman>   <acted_in>   <Eyes_Wide_Shut> .
<Tom_Cruise>      <acted_in>   <Eyes_Wide_Shut> .
<Nicole_Kidman>   <married_to> <Tom_Cruise> .
<Stanley_Kubrick> <directed>   <Eyes_Wide_Shut> .

Four triples, four nodes (Nicole_Kidman, Tom_Cruise, Eyes_Wide_Shut, Stanley_Kubrick), four directed labelled edges; e.g. the edge Nicole_Kidman → Tom_Cruise carries the label married_to. Likewise, <Albert_Einstein> <born_in> <Ulm> ., <Albert_Einstein> <occupation> <Physicist> ., <Ulm> <country> <Germany> . give 4 nodes and 3 edges.

With entity IRIs and literals:

wd:Q37459 wdt:P19    wd:Q18094 .                # Nicole Kidman, place of birth, Honolulu
wd:Q37459 wdt:P569   "1967-06-20"^^xsd:date .   # date of birth (typed literal)
wd:Q37459 rdfs:label "Nicole Kidman"@en .       # name (language-tagged literal)

Sources#

Slides 07, p. 5–6 (RDF, KG as graph), p. 7 (IRIs), pp. 8–9 (large KGs), p. 11 (interoperability)
DBIS_VL07 > Knowledge Graphs / RDF, DBIS_VL07 > IRIs, DBIS_VL07 > Interoperability

Connections

Mentioned in inline links7

Reification
An $n$ -ary fact relates $n > 2$ components, e.g. $\text{award}(\text{person}, \text{award}, \text{year}, \text{film})$ . RDF triples only express binary relations, so the fact is reified: introduce a fresh mediator entity $m$ for the fact a… §
SPARQL Query Structure
SPARQL (SPARQL Protocol And RDF Query Language) is the standard query language for RDF data (W3C standard, freely available). A basic query has the form §
SPARQL to SQL Translation
Let the triples be stored in one table triples(subject, predicate, object) and let SELECT ?v1 … ?vs WHERE { T1 . … . Tk } be a basic SPARQL query with $k$ query triples. The equivalent SQL query is built as follows: §
VL07 – Knowledge Graphs, SPARQL, SPARQL to SQL
The Resource Description Framework (RDF) is based on one principle: all data, whatever it is, is modelled as a set of triples (RDF Triples) §
RDF Triples — data model: all data as subject predicate object triples forming a directed knowledge graph §
SPARQL and Knowledge Graphs
RDF Triple — data model: all data as triples, a set of triples is a directed labelled graph §
Triples → patterns. A KG is a set of triples; a SPARQL query is a small graph of triples with variables, and its result table lists all matches in the KG. A variable shared by two triples is a join. §
VL01 – Overview, RDF & SPARQL, more SPARQLADIS
All data is a set of subject predicate object . triples (RDF Triple). Each distinct entity is a node and each triple a directed edge from subject to object, so a knowledge graph is a directed graph (Slides 8–9). §
RDF Triple — recap: the data model, the graph view, IRIs and interoperability (DBIS) §
Advanced SPARQLADIS
Basics from DBIS, used throughout: RDF Triple, SPARQL Query Structure (patterns, FILTER, LANG), Wikidata Prefix System, Ordering (ORDER BY, LIMIT). §

Search in these mentions →