The Resource Description Framework (RDF) is a framework for representing information in the Web. This document defines an abstract syntax (a data model) which serves to link all RDF-based languages and specifications. The abstract syntax has two key data structures:

RDF 1.2 introduces triple terms as another kind of RDF term which can be used as the object of another triple. RDF 1.2 also introduces directional language-tagged strings, which contain a base direction element that allows the initial text direction to be specified for presentation by a user agent.

RDF 1.2 Concepts introduces key concepts and terminology for RDF 1.2, discusses datatyping, and the handling of fragment identifiers in IRIs within RDF graphs.

This document is part of the RDF 1.2 document suite. It is the central RDF 1.2 specification and defines the core RDF concepts. Test suites and implementation reports of a number of RDF 1.2 specifications that build on this document are available through the [[[RDF11-TESTCASES]]] document [[RDF11-TESTCASES]].

RDF 1.2 Concepts is an update to [[RDF11-CONCEPTS]], which was itself, an update to [[RDF-CONCEPTS-20040210]].

Determine how to reference 1.2 test cases.


The Resource Description Framework (RDF) is a framework for representing information in the Web.

This document defines an abstract syntax (a data model) which serves to link all RDF-based languages and specifications, including:

Graph-based Data Model

The core structure of the abstract syntax is a set of triples, each consisting of a subject, a predicate and an object. A set of such triples is called an RDF graph. An RDF graph can be visualized as a node and directed-arc diagram, in which each triple is represented as a node-arc-node link.

An RDF graph with two nodes (Subject and Object) and a triple connecting them (Predicate)
An RDF graph with two nodes (Subject and Object) and a triple connecting them (Predicate)

There can be four kinds of nodes in an RDF graph: IRIs, literals, blank nodes, and triple terms.

Resources and Statements

Any IRI or literal denotes something in the world (the "universe of discourse"). These things are called resources. Anything can be a resource, including physical things, documents, abstract concepts, numbers and strings; the term is synonymous with "entity" as it is used in [[[RDF12-SEMANTICS]]] [[RDF12-SEMANTICS]]. The resource denoted by an IRI is called its referent, and the resource denoted by a literal is called its literal value. Literals have datatypes that define the range of possible values, such as strings, numbers, and dates. Special kinds of literals — language-tagged strings and directional language-tagged strings — respectively denote plain-text strings in a natural language, and plain-text strings in a natural language including an initial text direction.

Asserting an RDF triple says that some relationship, indicated by the predicate, holds between the resources denoted by the subject and object. This statement corresponding to an RDF triple is known as an RDF statement. The predicate itself is an IRI and denotes a property, that is, a resource that can be thought of as a binary relation. (Relations that involve more than two entities can only be indirectly expressed in RDF [[SWBP-N-ARYRELATIONS]].)

Unlike IRIs and literals, blank nodes do not identify specific resources. Statements involving blank nodes say that something with the given relationships exists, without explicitly naming it.

Quoted Triples

As the WG has moved away from the notion of quoted triples, this section needs to be changed such that it does not mention quoted triples anymore but, instead, describes whatever new concept the WG settles on (e.g., triple terms).

A quoted triple is an RDF term with the components of an RDF triple, which can be used as the subject or object of another triple.

A quoted triple is not necessarily asserted, allowing statements to be made about other statements that may not be asserted within an RDF graph. This allows statements to be made about relationships that may be contradictory. For a quoted triple to be asserted, it must also appear in a graph as an asserted triple.

The following diagram represents a statement with an unasserted quoted triple as the subject.

An RDF graph containing a triple that references an unasserted quoted triple (with grey background) as the subject
An RDF graph containing a triple that references an unasserted quoted triple (with grey background) as the subject.

A variation on the graph shown in can be described where the quoted triple is also asserted.

An RDF graph containing two triples where one of them contains the other as the subject; hence, that other triple is both quoted and asserted
An RDF graph containing two triples where one of them contains the other as the subject; hence, that other triple is both quoted and asserted.

Note that a quoted triple may also have a quoted triple as its subject or object.

This issue considers the potential for additional terminology related to quoted triples.

The Referent of an IRI

The resource denoted by an IRI is also called its referent. For some IRIs with particular meanings, such as those identifying XSD datatypes, the referent is fixed by this specification. For all other IRIs, what exactly is denoted by any given IRI is not defined by this specification. Other specifications may fix IRI referents, or apply other constraints on what may be the referent of any IRI.

Guidelines for determining the referent of an IRI are provided in other documents, like [[[WEBARCH]]] [[WEBARCH]] and [[[COOLURIS]]] [[COOLURIS]]. A very brief, informal, and partial account follows:

Perhaps the most important characteristic of IRIs in web architecture is that they can be dereferenced, and hence serve as starting points for interactions with a remote server. This specification is not concerned with such interactions. It does not define an interaction model. It only treats IRIs as globally unique identifiers in a graph data model that describes resources. However, those interactions are critical to the concept of [[[LINKED-DATA]]], [[LINKED-DATA]], which makes use of the RDF data model and serialization formats.

RDF Vocabularies and Namespace IRIs

An RDF vocabulary is a collection of IRIs intended for use in RDF graphs. For example, the IRIs documented in [[RDF12-SCHEMA]] are the RDF Schema vocabulary. RDF Schema can itself be used to define and document additional RDF vocabularies. Some such vocabularies are mentioned in the Primer [[RDF12-PRIMER]].

The IRIs in an RDF vocabulary often begin with a common substring known as a namespace IRI. Some namespace IRIs are associated by convention with a short name known as a namespace prefix. Some examples:

Some example namespace prefixes and IRIs
Namespace prefix Namespace IRI RDF vocabulary
rdf The RDF built-in vocabulary [[RDF-SYNTAX]]
rdfs The RDF Schema vocabulary [[RDF-SCHEMA]]
xsd The RDF-compatible XSD types

In some serialization formats it is common to abbreviate IRIs that start with namespace IRIs by using a namespace prefix in order to assist readability. For example, the IRI would be abbreviated as rdf:XMLLiteral. Note however that these abbreviations are not valid IRIs, and must not be used in contexts where IRIs are expected. Namespace IRIs and namespace prefixes are not a formal part of the RDF data model. They are merely a syntactic convenience for abbreviating IRIs.

The term “namespace” on its own does not have a well-defined meaning in the context of RDF, but is sometimes informally used to mean “namespace IRI” or “RDF vocabulary”.

RDF and Change over Time

The RDF data model is atemporal: RDF graphs are static snapshots of information.

However, RDF graphs can express information about events and about temporal aspects of other entities, given appropriate vocabulary terms.

Since RDF graphs are defined as mathematical sets, adding or removing triples from an RDF graph yields a different RDF graph.

We informally use the term RDF source to refer to a persistent yet mutable source or container of RDF graphs. An RDF source is a resource that may be said to have a state that can change over time. A snapshot of the state can be expressed as an RDF graph. For example, any web document that has an RDF-bearing representation may be considered an RDF source. Like all resources, RDF sources may be named with IRIs and therefore described in other RDF graphs.

Intuitively speaking, changes in the universe of discourse can be reflected in the following ways:

Working with Multiple RDF Graphs

As RDF graphs are sets of triples, they can be combined easily, supporting the use of data from multiple sources. Nevertheless, it is sometimes desirable to work with multiple RDF graphs while keeping their contents separate. RDF datasets support this requirement.

An RDF dataset is a collection of RDF graphs. All but one of these graphs have an associated IRI or blank node. They are called named graphs, and the IRI or blank node is called the graph name. The remaining graph does not have an associated IRI, and is called the default graph of the RDF dataset.

There are many possible uses for RDF datasets. One such use is to hold snapshots of multiple RDF sources.

Equivalence, Entailment and Inconsistency

An RDF triple encodes a statement—a simple logical expression, or claim about the world. An RDF graph is the conjunction (logical AND) of its triples. The precise details of this meaning of RDF triples and graphs are the subject of [[[RDF12-SEMANTICS]]] [[RDF12-SEMANTICS]], which yields the following relationships between RDF graphs:

An RDF graph A entails another RDF graph B if every possible arrangement of the world that makes A true also makes B true. When A entails B, if the truth of A is presumed or demonstrated then the truth of B is established.
Two RDF graphs A and B are equivalent if they make the same claim about the world. A is equivalent to B if and only if A entails B and B entails A.
An RDF graph is inconsistent if it contains an internal contradiction. There is no possible arrangement of the world that would make the expression true.

An entailment regime [[RDF12-SEMANTICS]] is a specification that defines precise conditions that make these relationships hold. RDF itself recognizes only some basic cases of entailment, equivalence and inconsistency. Other specifications, such as [[[RDF-SCHEMA]]] [[RDF-SCHEMA]] and OWL 2 [[OWL2-OVERVIEW]], add more powerful entailment regimes, as do some domain-specific vocabularies.

This specification does not constrain how implementations use the logical relationships defined by entailment regimes. Implementations may or may not detect inconsistencies, and may make all, some or no entailed information available to users.

RDF Documents and Syntaxes

An RDF document is a document that encodes an RDF graph or RDF dataset in a concrete RDF syntax, such as Turtle [[RDF12-TURTLE]], RDFa [[RDFA-CORE]], JSON-LD [[JSON-LD11]], or TriG [[RDF12-TRIG]]. RDF documents enable the exchange of RDF graphs and RDF datasets between systems.

A concrete RDF syntax may offer many different ways to encode the same RDF graph or RDF dataset, for example through the use of namespace prefixes, IRI references, blank node identifiers, and different ordering of triples. While these aspects can have great effect on the convenience of working with the RDF document, they are not significant for its meaning.

This specification, RDF 1.2 Concepts and Abstract Syntax, defines a data model and related terminology for use in other specifications, such as concrete RDF syntaxes, API specifications, and query languages. Implementations cannot directly conform to RDF 1.2 Concepts and Abstract Syntax, but can conform to such other specifications that normatively reference terms defined here.

This specification establishes two conformance levels:

The conformance levels described above are tentative, and still the subject of group discussion. An alternative to conformance levels, "profiles", may be adopted instead, abandoned, or described in another specification.

Strings in RDF

RDF uses Unicode [[Unicode]] as the fundamental representation for string values. Within this, and related specifications, the term string, or RDF string, is used to describe an ordered sequence of zero or more Unicode code points which are Unicode scalar values. Unicode scalar values do not include the surrogate code points. Note that most concrete RDF syntaxes require the use of the UTF-8 character encoding [[!RFC3629]], and use the `\u0000` or `\U00000000` forms to express certain non-character values.

A string is identical to another string if it consists of the same sequence of code points. An implementation MAY determine string equality by comparing the code units of two strings that use the same Unicode character encoding (UTF-8 or UTF-16) without decoding the string into a Unicode code point sequence.

RDF Graphs

An RDF graph is a set of RDF triples.

Do we still need to define asserted triple?.


An RDF triple (usually called "triple") is a 3-tuple (|s|, |p|, |o|) where:

A triple term is a 3-tuple that is defined recursively as follows:

Given a triple (|s|, |p|, |o|), |s| is called the subject of the triple, |p| is called the predicate of the triple, and |o| is called the object of the triple. Similarly, given a triple term (|s|, |p|, |o|), |s| is called the subject of the triple term, |p| is called the predicate of the triple term, and |o| is called the object of the triple term.

While, syntactically, the notion of an RDF triple and the notion of a triple term are the same, they represent different concepts. RDF triples are the members of RDF graphs, whereas triple terms can be used as components of RDF triples.

The definition of triple term is recursive. That is, a triple term can itself have an object component which is another triple term. However, by this definition, cycles of triple terms cannot be created.

IRIs, literals, blank nodes, and triple terms are collectively known as RDF terms.

IRIs, literals, blank nodes, and triple terms are distinct and distinguishable. For example, a literal with the string as its lexical form is not equal to the IRI, nor to a blank node with the blank node identifier

The set of nodes of an RDF graph is the set of subjects and objects of the triples in the graph. It is possible for a predicate IRI to also occur as a node in the same graph.


An IRI (Internationalized Resource Identifier) within an RDF graph is a string that conforms to the syntax defined in RFC 3987 [[!RFC3987]].

For convenience, a complete [[ABNF]] grammar from [[RFC3987]] is provided in .

IRIs in the RDF abstract syntax MUST be resolved per [[RFC3986]], and MAY contain a fragment identifier.

IRI equality: Two IRIs are the same if and only if they consist of the same sequence of Unicode code points, as in Simple String Comparison in section 5.3.1 of [[!RFC3987]]. (This is done in the abstract syntax, so the IRIs are resolved IRIs with no escaping or encoding.) Further normalization MUST NOT be performed before this comparison.

URIs and IRIs: IRIs are a generalization of URIs [[RFC3986]] that permits a wider range of Unicode characters [[UNICODE]]. Every URI and URL is an IRI, but not every IRI is an URI. In RDF, IRIs are used as IRI references, as defined in [[RFC3987]] section 1.3. An IRI reference is common usage of an Internationalized Resource Identifier. An IRI reference refers to either a resolved IRI or relative IRI reference, as described by the IRI-reference production in . The abstract syntax uses only fully resolved IRIs. When IRIs are used in operations that are only defined for URIs, they must first be converted according to the mapping defined in section 3.1 of [[!RFC3987]]. A notable example is retrieval over the HTTP protocol. The mapping involves UTF-8 encoding of non-ASCII characters, %-encoding of octets not allowed in URIs, and Punycode-encoding of domain names.

URLs: The [[[URL]]] is largely compatible with [[RFC3987]] IRIs, but is based on a processing model important for implementation within web browsers and are not described using an [[ABNF]] grammar.

Relative IRI references: Some concrete RDF syntaxes permit relative IRI references as a convenient shorthand that allows authoring of documents independently from their final publishing location. Relative IRI references must be resolved against a base IRI. Therefore, the RDF graph serialized in such syntaxes is well-defined only if a base IRI can be established [[RFC3986]].

IRI normalization: Interoperability problems can be avoided by minting only IRIs that are normalized according to Section 5 of [[!RFC3987]]. Non-normalized forms that are best avoided include:

  • Uppercase characters in scheme names and domain names
  • Percent-encoding of characters where it is not required by IRI syntax
  • Explicitly stated HTTP default port (; is preferable
  • Completely empty path in HTTP IRIs (; is preferable
  • /./” or “/../” in the path component of an IRI
  • Lowercase hexadecimal letters within percent-encoding triplets (“%3F” is preferable over “%3f”)
  • Punycode-encoding of Internationalized Domain Names in IRIs [[RFC3492]]
  • IRIs that are not in Unicode Normalization Form C [[UAX15]]


Literals are used for values such as strings, numbers, and dates.

A literal in an RDF graph consists of two, three, or four elements:

A literal is a language-tagged string if the third element is present and the fourth element is not present. Lexical representations of language tags MAY be case normalized, (for example, by converting to lower case).

A literal is a directional language-tagged string if both the third element and fourth elements are present. The third element, the language tag, is treated identically as in a language-tagged string, and the fourth element, base direction, MUST be either `ltr` or `rtl`, which MUST be in lower case.

Please note that concrete syntaxes MAY support simple literals consisting of only a lexical form without any datatype IRI, language tag, or base direction. Simple literals are syntactic sugar for abstract syntax literals with the datatype IRI (which is commonly abbreviated as xsd:string). Similarly, most concrete syntaxes represent language-tagged strings and directional language-tagged strings without the datatype IRI because it always equals either (rdf:langString) or (rdf:dirLangString), respectively.

The literal value associated with a literal is:

  1. If the literal is a language-tagged string, then the literal value is a pair consisting of its lexical form and its language tag, in that order.
  2. if the literal is a directional language-tagged string, then the literal value is a tuple of its lexical form, its language tag, and its base direction, likewise in that order.
  3. If the literal's datatype IRI is in the set of recognized datatype IRIs, let d be the referent of the datatype IRI.
    1. If the literal's lexical form is in the lexical space of d, then the literal value is the result of applying the lexical-to-value mapping of d to the lexical form.
    2. Otherwise, the literal is ill-typed and no literal value can be associated with the literal. Such a case produces a semantic inconsistency but is not syntactically ill-formed. Implementations MUST accept ill-typed literals and produce RDF graphs from them. Implementations MAY produce warnings when encountering ill-typed literals.
  4. If the literal's datatype IRI is not in the set of recognized datatype IRIs, then the literal value is not defined by this specification.

Literal term equality: Two literals are term-equal (the same RDF literal) if and only if the two lexical forms, the two datatype IRIs, the two language tags (if any), and the two base directions (if any) compare equal, using case sensitive matching (see description of string comparison in ). Thus, two literals can have the same value without being the same RDF term. For example:


denote the same value, but are not the same literal RDF terms and are not term-equal because their lexical forms differ.

Initial Text Direction

The base direction of a directional language-tagged string provides a means of establishing the initial direction of text, including text which is a mixture of right-to-left and left-to-right scripts. The [[[?UAX9]]] [[?UAX9]] provides support for automatically rendering a sequence of characters in logical order, so that they are visually ordered as expected, but this is not sufficient to correctly render bidirectional text.

For example, some text with a language tag "`he`" might be displayed in a left-to-right context (such as an English web page) as פעילות הבינאום, W3C, which is incorrect. When provided to a user agent including base direction information (such as using HTML's `dir` attribute) it can then be correctly presented as:

פעילות הבינאום, W3C

Blank Nodes

Blank nodes are disjoint from IRIs and literals. Otherwise, the set of possible blank nodes is arbitrary. RDF makes no reference to any internal structure of blank nodes.

Blank node identifiers are local identifiers that are used in some concrete RDF syntaxes or RDF store implementations. They are always locally scoped to the file or RDF store, and are not persistent or portable identifiers for blank nodes. Blank node identifiers are not part of the RDF abstract syntax, but are entirely dependent on the concrete syntax or implementation. The syntactic restrictions on blank node identifiers, if any, therefore also depend on the concrete RDF syntax or implementation. Implementations that handle blank node identifiers in concrete syntaxes need to be careful not to create the same blank node from multiple occurrences of the same blank node identifier except in situations where this is supported by the syntax.

The term "blank node label" is sometimes used informally as an alternative to the term blank node identifier. This alternative was also used in earlier versions of some RDF-related specifications such as [[SPARQL11-QUERY]]. In the interest of consistency, the use of this alternative term is discouraged now.

Replacing Blank Nodes with IRIs

Blank nodes do not have identifiers in the RDF abstract syntax. The blank node identifiers introduced by some concrete syntaxes have only local scope and are purely an artifact of the serialization.

In situations where stronger identification is needed, systems MAY systematically replace some or all of the blank nodes in an RDF graph with IRIs. Systems wishing to do this SHOULD mint a new, globally unique IRI (a Skolem IRI) for each blank node so replaced.

This transformation does not appreciably change the meaning of an RDF graph, provided that the Skolem IRIs do not occur anywhere else. It does however permit the possibility of other graphs subsequently using the Skolem IRIs, which is not possible for blank nodes.

Systems may wish to mint Skolem IRIs in such a way that they can recognize the IRIs as having been introduced solely to replace blank nodes. This allows a system to map IRIs back to blank nodes if needed.

Systems that want Skolem IRIs to be recognizable outside of the system boundaries SHOULD use a well-known IRI [[RFC5785]] with the registered name genid. This is an IRI that uses the HTTP or HTTPS scheme, or another scheme that has been specified to use well-known IRIs; and whose path component starts with /.well-known/genid/.

For example, the authority responsible for the domain could mint the following recognizable Skolem IRI:

RFC 5785 [[RFC5785]] only specifies well-known URIs, not IRIs. For the purpose of this document, a well-known IRI is any IRI that results in a well-known URI after IRI-to-URI mapping [[!RFC3987]].

Graph Comparison

Two RDF graphs G and G' are isomorphic (that is, they have an identical form) if there is a bijection M between the sets of nodes of the two graphs, such that:

  1. M maps blank nodes to blank nodes.
  2. M(lit)=lit for all RDF literals lit which are nodes of G.
  3. M(iri)=iri for all IRIs iri which are nodes of G.
  4. The triple ( s, p, o ) is in G if and only if the triple ( M(s), p, M(o) ) is in G'

See also: IRI equality, literal term equality.

With this definition, M shows how each blank node in G can be replaced with a new blank node to give G'. Graph isomorphism is needed to support the RDF Test Cases [[RDF11-TESTCASES]] specification.

RDF Datasets

An RDF dataset is a collection of RDF graphs, and comprises:

Blank nodes can be shared between graphs in an RDF dataset.

Despite the use of the word “name” in “named graph”, the graph name is not required to denote the graph. It is merely syntactically paired with the graph. RDF does not place any formal restrictions on what resource the graph name may denote, nor on the relationship between that resource and the graph. A discussion of different RDF dataset semantics can be found in [[RDF11-DATASETS]].

Some RDF dataset implementations do not track empty named graphs. Applications can avoid interoperability issues by not ascribing importance to the presence or absence of empty named graphs.

SPARQL 1.2 [[SPARQL12-CONCEPTS]] also defines the concept of an RDF Dataset. The definition of an RDF Dataset in SPARQL 1.1 and this specification differ slightly in that this specification allows RDF Graphs to be identified using either an IRI or a blank node. SPARQL 1.1 Query Language only allows RDF Graphs to be identified using an IRI. Existing SPARQL implementations might not allow blank nodes to be used to identify RDF Graphs for some time, so their use can cause interoperability problems. Skolemizing blank nodes used as graph names can be used to overcome these interoperability problems.

RDF Dataset Comparison

Two RDF datasets (the RDF dataset D1 with default graph DG1 and any named graph NG1 and the RDF dataset D2 with default graph DG2 and any named graph NG2) are dataset-isomorphic if and only if there is a bijection M between the nodes, triples and graphs in D1 and those in D2 such that:

  1. M maps blank nodes to blank nodes;
  2. M is the identity map on literals and URIs;
  3. For every triple <s p o>, M(<s, p, o>)= <M(s), M(p), M(o)>;
  4. For every graph G={t1, ..., tn}, M(G)={M(t1), ..., M(tn)};
  5. DG2 = M(DG1); and
  6. <n, G> is in NG1 if and only if <M(n), M(G)> is in NG2.

Content Negotiation of RDF Datasets

Web resources may have multiple representations that are made available via content negotiation [[WEBARCH]]. A representation may be returned in an RDF serialization format that supports the expression of both RDF datasets and RDF graphs. If an RDF dataset is returned and the consumer is expecting an RDF graph, the consumer is expected to use the RDF dataset's default graph.

Dataset as a Set of Quads

A quad is a triple associated with an optional graph name and is used when referring to triples within an RDF dataset.

A quad can be represented as a tuple composed of subject, predicate, object, and an optional graph name.

An RDF dataset can be considered to be a set of quads where quads with no graph name supply the triples of the default graph, and quads with the same graph name supply the triples of the named graph with that name.

Although a quad without a graph name consists of the same three components as a triple, it is a distinct concept, as it specifically captures the notion of a triple within the default graph of an RDF dataset.


Datatypes are used with RDF literals to represent values such as strings, numbers and dates. The datatype abstraction used in RDF is compatible with XML Schema [[!XMLSCHEMA11-2]]. Any datatype definition that conforms to this abstraction MAY be used in RDF, even if not defined in terms of XML Schema. RDF re-uses many of the XML Schema built-in datatypes, and defines three additional datatypes, rdf:JSON, rdf:HTML, and rdf:XMLLiteral. The list of datatypes supported by an implementation is determined by its recognized datatype IRIs.

A datatype consists of a lexical space, a value space and a lexical-to-value mapping, and is denoted by one or more IRIs.

The lexical space of a datatype is a set of strings.

The lexical-to-value mapping of a datatype is a set of pairs whose first element belongs to the lexical space, and the second element belongs to the value space of the datatype. Each member of the lexical space is paired with exactly one value, and is a lexical representation of that value. The mapping can be seen as a function from the lexical space to the value space.

Language-tagged strings have the datatype IRI (commonly abbreviated as rdf:langString). No datatype is formally defined for this IRI because the definition of datatypes does not accommodate language tags in the lexical space. The value space associated with this datatype IRI is the set of all pairs of strings and language tags.

For example, the XML Schema datatype xsd:boolean, where each member of the value space has two lexical representations, is defined as follows:

Lexical space:
{“true”, “false”, “1”, “0”}
Value space:
{true, false}
Lexical-to-value mapping
{ <“true”, true>, <“false”, false>, <“1”, true>, <“0”, false>, }

The literals that can be defined using this datatype are:

This table lists the literals of type xsd:boolean.
Literal Value
<“true”, xsd:boolean> true
<“false”, xsd:boolean> false
<“1”, xsd:boolean> true
<“0”, xsd:boolean> false

The XML Schema Built-in Datatypes

IRIs of the form, where xxx is the name of a datatype, denote the built-in datatypes defined in [[[XMLSCHEMA11-2]]] [[!XMLSCHEMA11-2]]. The XML Schema built-in types listed in the following table are the RDF-compatible XSD types. Their use is RECOMMENDED.

Readers might note that the only safe datatypes for transferring binary information are `xsd:hexBinary` and `xsd:base64Binary`.

A list of the RDF-compatible XSD types, with short descriptions
DatatypeValue space (informative)
Core typesxsd:stringCharacter strings
xsd:booleantrue, false
xsd:decimalArbitrary-precision decimal numbers
xsd:integerArbitrary-size integer numbers
IEEE floating-point
xsd:double64-bit floating point numbers incl. ±Inf, ±0, NaN
xsd:float32-bit floating point numbers incl. ±Inf, ±0, NaN
Time and date xsd:dateDates (yyyy-mm-dd) with or without timezone
xsd:timeTimes (hh:mm:ss.sss…) with or without timezone
xsd:dateTimeDate and time with or without timezone
xsd:dateTimeStampDate and time with required timezone
Recurring and
partial dates
xsd:gYearGregorian calendar year
xsd:gMonthGregorian calendar month
xsd:gDayGregorian calendar day of the month
xsd:gYearMonthGregorian calendar year and month
xsd:gMonthDayGregorian calendar month and day
xsd:durationDuration of time
xsd:yearMonthDurationDuration of time (months and years only)
xsd:dayTimeDurationDuration of time (days, hours, minutes, seconds only)
integer numbers
xsd:byte-128…+127 (8 bit)
xsd:short-32768…+32767 (16 bit)
xsd:int-2147483648…+2147483647 (32 bit)
xsd:long-9223372036854775808…+9223372036854775807 (64 bit)
xsd:unsignedByte0…255 (8 bit)
xsd:unsignedShort0…65535 (16 bit)
xsd:unsignedInt0…4294967295 (32 bit)
xsd:unsignedLong0…18446744073709551615 (64 bit)
xsd:positiveIntegerInteger numbers >0
xsd:nonNegativeIntegerInteger numbers ≥0
xsd:negativeIntegerInteger numbers <0
xsd:nonPositiveIntegerInteger numbers ≤0
Encoded binary data xsd:hexBinaryHex-encoded binary data
xsd:base64BinaryBase64-encoded binary data
XSD types
xsd:anyURIResolved or relative URI and IRI references
xsd:languageLanguage tags per [[BCP47]]
xsd:normalizedStringWhitespace-normalized strings
xsd:tokenTokenized strings
xsd:NameXML Names
xsd:NCNameXML NCNames

The other built-in XML Schema datatypes are unsuitable for various reasons and SHOULD NOT be used:

The value spaces of xsd:double and xsd:float do not include all decimal numbers. For every literal of either of these two datatypes, the value of the literal is a value that can be represented as an IEEE 754-2008 binary floating point representation of the corresponding precision. For instance, the literal with lexical form `"0.1"` and datatype xsd:float denotes the number `0.100000001490116119384765625`. Rather than xsd:double or xsd:float, the datatype xsd:decimal can be used to accurately capture arbitrary decimal numbers.

Datatype IRIs

Datatypes are identified by IRIs. If D is a set of IRIs which are used to refer to datatypes, then the elements of D are called recognized datatype IRIs. Recognized IRIs have fixed referents. If any IRI of the form is recognized, it MUST refer to the RDF-compatible XSD type named xsd:xxx for every XSD type listed in section 5.1.

The datatypes identified by the three IRIs below are defined in Appendix :

RDF processors are not required to recognize datatype IRIs. Any literal typed with an unrecognized IRI is treated just like an unknown IRI, i.e. as referring to an unknown thing. Applications MAY give a warning message if they are unable to determine the referent of an IRI used in a typed literal, but they SHOULD NOT reject such RDF as either a syntactic or semantic error.

Other specifications MAY impose additional constraints on datatype IRIs, for example, require support for certain datatypes.

Semantic extensions of RDF might choose to recognize other datatype IRIs and require each of them to refer to a fixed datatype. See [[[RDF12-SEMANTICS]]] [[RDF12-SEMANTICS]] for more information on semantic extensions.

The Web Ontology Language [[OWL2-OVERVIEW]] offers facilities for formally defining custom datatypes that can be used with RDF. Furthermore, a practice for identifying user-defined simple XML Schema datatypes is suggested in [[SWBP-XSCH-DATATYPES]]. RDF implementations are not required to support either of these facilities.

Fragment Identifiers

RDF uses IRIs, which may include fragment identifiers, as resource identifiers. The semantics of fragment identifiers is defined in RFC 3986 [[RFC3986]]: They identify a secondary resource that is usually a part of, a view of, defined in, or described in the primary resource, and the precise semantics depend on the set of representations that might result from a retrieval action on the primary resource.

This section discusses the handling of fragment identifiers in representations that encode RDF graphs.

In RDF-bearing representations of a primary resource, e.g., <>, the secondary resource identified by a fragment identifier, e.g., bar, is the resource denoted by the full IRI in the RDF graph, which would be <> in this case. Since IRIs in RDF graphs can denote anything, this can be something external to the representation, or even external to the web.

In this way, the RDF-bearing representation acts as an intermediary between the web-accessible primary resource, and some set of possibly non-web or abstract entities that the RDF graph may describe.

In cases where other specifications constrain the semantics of fragment identifiers in RDF-bearing representations, the encoded RDF graph should use fragment identifiers in a way that is consistent with these constraints. For example, in an HTML+RDFa document [[HTML-RDFA]], a fragment identifier such as chapter1 may identify a document section via the semantics of HTML's @name or @id attributes. Such an IRI, e.g., <#chapter1>, should then be taken to denote that same section in any RDFa-encoded triples within the same document. Similarly, fragment identifiers should be used consistently in resources with multiple representations that are made available via content negotiation [[WEBARCH]]. For example, if the fragment identifier chapter1 identifies a document section in an HTML representation of the primary resource, then the IRI <#chapter1> should be taken to denote that same section in all RDF-bearing representations of the same primary resource.

Generalized RDF Triples, Graphs, and Datasets

It is sometimes convenient to loosen the requirements on RDF triples. For example, the completeness of the RDFS entailment rules is easier to show with a generalization of RDF triples.

A generalized RDF triple is a triple having a subject, a predicate, and an object, where each can be an IRI, a blank node or a literal. A generalized RDF graph is a set of generalized RDF triples. A generalized RDF dataset comprises a distinguished generalized RDF graph, and zero or more pairs each associating an IRI, a blank node, or a literal to a generalized RDF graph.

Generalized RDF triples, graphs, and datasets differ from standard normative RDF triples, graphs, and datasets only by allowing IRIs, blank nodes, and literals to appear in any position, i.e., as subject, predicate, object, or graph name.

Any user of generalized RDF triples, graphs, or datasets needs to be aware that these notions are non-standard extensions of RDF, and their use may cause interoperability problems. There is no requirement for any RDF tool to accept, process, or produce anything beyond standard normative RDF triples, graphs, and datasets.

Additional Datatypes

This section defines additional datatypes that RDF processors MAY support.

The rdf:HTML Datatype

RDF provides for HTML content as a possible literal value. This allows markup in literal values. Such content is indicated in an RDF graph using a literal whose datatype is set to rdf:HTML.

The rdf:HTML datatype is defined as follows:

The IRI denoting this datatype
The value space
is the set of DOM DocumentFragment nodes [[DOM]]. Two DocumentFragment nodes node and otherNode are considered equal if and only if the DOM method node.{{Node/isEqualNode(otherNode)}} [[DOM]] returns true.
The lexical-to-value mapping

Each member of the lexical space is associated with the result of applying the following algorithm:

  • Let domnodes be the list of DOM nodes [[DOM]] that result from applying the HTML fragment parsing algorithm [[HTML5]] to the input string, without a context element.
  • Let domfrag be a DOM DocumentFragment [[DOM]] whose childNodes attribute is equal to domnodes
  • Return domfrag.{{Node/normalize()}}

Any language annotation (lang="…") or XML namespaces (xmlns) desired in the HTML content must be included explicitly in the HTML literal. Relative URLs in attributes such as href do not have a well-defined base URL and are best avoided. RDF applications may use additional equivalence relations, such as that which relates an xsd:string with an rdf:HTML literal corresponding to a single text node of the same string.

The rdf:XMLLiteral Datatype

RDF provides for XML content as a possible literal value. Such content is indicated in an RDF graph using a literal whose datatype is set to rdf:XMLLiteral.

The rdf:XMLLiteral datatype is defined as follows:

The IRI denoting this datatype
The lexical space
is the set of all strings which are well-balanced, self-contained XML content [[XML11]]; and for which embedding between an arbitrary XML start tag and an end tag yields a document conforming to [[[XML-NAMES]]] [[XML-NAMES]].
The value space
is the set of DOM DocumentFragment nodes [[DOM]]. Two DocumentFragment nodes node and otherNode are considered equal if and only if the DOM method node.{{Node/isEqualNode(otherNode)}} returns true.
The lexical-to-value mapping

Each member of the lexical space is associated with the result of applying the following algorithm:

  • Let domfrag be a DOM DocumentFragment node [[DOM]] corresponding to the input string
  • Return domfrag.{{Node/normalize()}}
The canonical mapping
defines a canonical lexical form [[XMLSCHEMA11-2]] for each member of the value space. The rdf:XMLLiteral canonical mapping is the exclusive XML canonicalization method (with comments, with empty InclusiveNamespaces PrefixList) [[XML-EXC-C14N]].

Any XML namespace declarations (xmlns), language annotation (xml:lang) or base URI declarations (xml:base) desired in the XML content must be included explicitly in the XML literal. Note that some concrete RDF syntaxes may define mechanisms for inheriting them from the context (e.g., @parseType="literal" in RDF/XML [[RDF12-XML]].

The rdf:JSON Datatype

RDF provides for JSON content as a possible literal value. This includes allowing markup in literal values. Such content is indicated in an RDF graph as a literal whose datatype is set to rdf:JSON.

The rdf:JSON datatype is defined as follows:

The IRI denoting this datatype
The lexical space
is the set of all RDF strings that conform to the JSON Grammar as described in Section 2 JSON Grammar of [[RFC8259]].
The value space
is the set of all RDF strings that conform to the JSON Grammar as described in Section 2 JSON Grammar of [[RFC8259]], and furthermore comply with the following constraints:
  • They MUST NOT contain any unnecessary whitespace.
  • Keys in objects MUST be ordered lexicographically.
  • Native Numeric values MUST be serialized according to Section of [[ECMASCRIPT]].
  • Strings MUST be serialized with Unicode code points from U+0000 through U+001F using lower case hexadecimal Unicode notation (`\uhhhh`) except for the set of predefined JSON control characters — `U+0008`, `U+0009`, `U+000A`, `U+000C`, and `U+000D` — which SHOULD be serialized as `\b`, `\t`, `\n`, `\f`, and `\r`, respectively. All other Unicode characters SHOULD be serialized "as is", except `U+005C` (`\`) and `U+0022` (`"`), which SHOULD be serialized as `\\` and `\"`, respectively.
The JSON Canonicalization Scheme (JCS) [[RFC8785]] is an emerging standard for JSON canonicalization. This rdf:JSON specification will likely be updated to require such a canonical representation. Users are cautioned against depending on the lexical representation of literals with the rdf:JSON datatype as RDF literals, as the specifics of serialization may change in a future revision of this document.
Despite being defined as a set of strings, this value space is considered distinct from the value space of xsd:string, in order to avoid side effects with existing specifications.
The lexical-to-value mapping
maps every element of the lexical space to the result of
  1. parsing it into an internal representation consistent with the [[ECMASCRIPT]] representation created by using the JSON.parse function as defined in Section 24.5 The JSON Object of [[ECMASCRIPT]],
  2. then serializing it into the JSON format [[RFC8259]] in compliance with the constraints of the value space described above.
The canonical mapping
maps any element of the value space to the identical string in the lexical space.

Privacy Considerations

RDF is used to express arbitrary application data, which may include the expression of personally identifiable information (PII) or other information which could be considered sensitive. Authors publishing such information are advised to carefully consider the needs and use of publishing such information, as well as the applicable regulations for the regions where the data is expected to be consumed and potentially revealed (e.g., GDPR, CCPA, others), particularly whether authorization measures are needed for access to the data.

Security Considerations

The RDF Abstract Syntax is not used directly for conveying information, although concrete serialization forms are specifically intended to do so.

Applications MAY evaluate given data to infer more assertions or to dereference IRIs, invoking the security considerations of the scheme for that IRI. Note in particular, the privacy issues in [[RFC3023]] section 10 for HTTP IRIs. Data obtained from an inaccurate or malicious data source may lead to inaccurate or misleading conclusions, as well as the dereferencing of unintended IRIs. Care must be taken to align the trust in consulted resources with the sensitivity of the intended use of the data; inferences of potential medical treatments would likely require different trust than inferences for trip planning.

RDF is used to express arbitrary application data; security considerations will vary by domain of use. Security tools and protocols applicable to text (for example, PGP encryption, checksum validation, password-protected compression) may also be used on RDF documents. Security/privacy protocols must be imposed which reflect the sensitivity of the embedded information.

RDF can express data which is presented to the user, such as RDF Schema labels. Applications rendering strings retrieved from untrusted RDF documents, or using unescaped characters, SHOULD use warnings and other appropriate means to limit the possibility that malignant strings might be used to mislead the reader. The security considerations in the media type registration for XML ([[RFC3023]] section 10) provide additional guidance around the expression of arbitrary data and markup.

RDF uses IRIs as term identifiers. Applications interpreting data expressed in RDF SHOULD address the security issues of [[[RFC3987]]] [[RFC3987]] Section 8, as well as [[[RFC3986]]] [[RFC3986]] Section 7.

Multiple IRIs may have the same appearance. Characters in different scripts may look similar (for instance, a Cyrillic "о" may appear similar to a Latin "o"). A character followed by combining characters may have the same visual representation as another character (for example, LATIN SMALL LETTER "E" followed by COMBINING ACUTE ACCENT has the same visual representation as LATIN SMALL LETTER "E" WITH ACUTE). Any person or application that is writing or interpreting data in RDF must take care to use the IRI that matches the intended semantics, and avoid IRIs that may look similar. Further information about matching visually similar characters can be found in [[[UNICODE-SECURITY]]] [[UNICODE-SECURITY]] and [[[RFC3987]]] [[RFC3987]] Section 8.

These considerations are a more generic form of Security Considerations for [[RDF12-TURTLE]], [[RDF12-TRIG]], [[RDF12-N-TRIPLES]], and [[RDF12-N-QUADS]].

Internationalization Considerations

Unicode [[UNICODE]] provides a mechanism for signaling direction within a string (see [[[UAX9]]] [[UAX9]]). RDF provides a mechanism for specifying the base direction of a directional language-tagged string to signal the initial text direction of a string. For most human language strings, but particularly for those whose base direction cannot be accurately determined from the string content, is it valuable to have an external indicator in order to get the proper display and isolation of the value. One example of such an indicator is the [[HTML]] dir attribute. see [[STRING-META]].

[[[JSON-LD11]]] [[JSON-LD11]] introduced the i18n namespace to use a datatype to specify both the base direction an language tag of an RDF literal.

IRI Grammar

The following [[ABNF]] grammar applies the changes from [[RFC3987]] and [[RFC6874]] to the section Collected ABNF for URI of [[RFC3986]] to give a consolidated grammar for IRIs.

This is provided for convenience only. If it differs from definitions in [[RFC3986]], [[RFC3987]], or any subsequent updates, then those definitions should be used.


The ABNF can also be accessed directly from iri-grammar.abnf.


Acknowledgments for RDF 1.0

The editors of the original version of the spec were Graham Klyne (Nine by Nine) and Jeremy J. Carroll (Hewlett Packard Labs).

This document contains a significant contribution from Pat Hayes, Sergey Melnik and Patrick Stickler, under whose leadership was developed the framework described in the RDF family of specifications for representing datatyped values, such as integers and dates.

The editors acknowledge valuable contributions from the following: Frank Manola, Pat Hayes, Dan Brickley, Jos de Roo, Dave Beckett, Patrick Stickler, Peter F. Patel-Schneider, Jerome Euzenat, Massimo Marchiori, Tim Berners-Lee, Dave Reynolds and Dan Connolly.

Jeremy Carroll thanks Oreste Signore, his host at the W3C Office in Italy and Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo", part of the Consiglio Nazionale delle Ricerche, where Jeremy is a visiting researcher.

This document is a product of extended deliberations by the RDFcore Working Group, whose members have included: Art Barstow (W3C), Dave Beckett (ILRT), Dan Brickley (ILRT), Dan Connolly (W3C), Jeremy Carroll (Hewlett Packard), Ron Daniel (Interwoven Inc), Bill dehOra (InterX), Jos De Roo (AGFA), Jan Grant (ILRT), Graham Klyne (Nine by Nine), Frank Manola (MITRE Corporation), Brian McBride (Hewlett Packard), Eric Miller (W3C), Stephen Petschulat (IBM), Patrick Stickler (Nokia), Aaron Swartz (HWG), Mike Dean (BBN Technologies / Verizon), R. V. Guha (Alpiri Inc), Pat Hayes (IHMC), Sergey Melnik (Stanford University) and Martyn Horner (Profium Ltd).

This specification also draws upon an earlier RDF Model and Syntax document edited by Ora Lassilla and Ralph Swick, and RDF Schema edited by Dan Brickley and R. V. Guha. RDF and RDF Schema Working Group members who contributed to this earlier work are: Nick Arnett (Verity), Tim Berners-Lee (W3C), Tim Bray (Textuality), Dan Brickley (ILRT / University of Bristol), Walter Chang (Adobe), Sailesh Chutani (Oracle), Dan Connolly (W3C), Ron Daniel (DATAFUSION), Charles Frankston (Microsoft), Patrick Gannon (CommerceNet), R. V. Guha (Epinions, previously of Netscape Communications), Tom Hill (Apple Computer), Arthur van Hoff (Marimba), Renato Iannella (DSTC), Sandeep Jain (Oracle), Kevin Jones, (InterMind), Emiko Kezuka (Digital Vision Laboratories), Joe Lapp (webMethods Inc.), Ora Lassila (Nokia Research Center), Andrew Layman (Microsoft), Ralph LeVan (OCLC), John McCarthy (Lawrence Berkeley National Laboratory), Chris McConnell (Microsoft), Murray Maloney (Grif), Michael Mealling (Network Solutions), Norbert Mikula (DataChannel), Eric Miller (OCLC), Jim Miller (W3C, emeritus), Frank Olken (Lawrence Berkeley National Laboratory), Jean Paoli (Microsoft), Sri Raghavan (Digital/Compaq), Lisa Rein (webMethods Inc.), Paul Resnick (University of Michigan), Bill Roberts (KnowledgeCite), i Tsuyoshi Sakata (Digital Vision Laboratories), Bob Schloss (IBM), Leon Shklar (Pencom Web Works), David Singer (IBM), Wei (William) Song (SISU), Neel Sundaresan (IBM), Ralph Swick (W3C), Naohiko Uramoto (IBM), Charles Wicksteed (Reuters Ltd.), Misha Wolf (Reuters Ltd.) and Lauren Wood (SoftQuad).

Acknowledgments for RDF 1.1

The editors of the RDF 1.1 version of the spec were Richard Cyganiak (DERI), David Wood (3 Round Stones), and Markus Lanthaler (Graz University of Technology).

The editors acknowledge valuable contributions from Thomas Baker, Tim Berners-Lee, David Booth, Dan Brickley, Gavin Carothers, Jeremy Carroll, Pierre-Antoine Champin, Dan Connolly, John Cowan, Martin J. Dürst, Alex Hall, Steve Harris, Sandro Hawke, Pat Hayes, Ivan Herman, Peter F. Patel-Schneider, Addison Phillips, Eric Prud'hommeaux, Nathan Rixham, Andy Seaborne, Leif Halvard Silli, Guus Schreiber, Dominik Tomaszuk, and Antoine Zimmermann.

The membership of the RDF Working Group included Thomas Baker, Scott Bauer, Dan Brickley, Gavin Carothers, Pierre-Antoine Champin, Olivier Corby, Richard Cyganiak, Souripriya Das, Ian Davis, Lee Feigenbaum, Fabien Gandon, Charles Greer, Alex Hall, Steve Harris, Sandro Hawke, Pat Hayes, Ivan Herman, Nicholas Humfrey, Kingsley Idehen, Gregg Kellogg, Markus Lanthaler, Arnaud Le Hors, Peter F. Patel-Schneider, Eric Prud'hommeaux, Yves Raimond, Nathan Rixham, Guus Schreiber, Andy Seaborne, Manu Sporny, Thomas Steiner, Ted Thibodeau, Mischa Tuffield, William Waites, Jan Wielemaker, David Wood, Zhe Wu, and Antoine Zimmermann.

Acknowledgments for RDF 1.2

In addition to the editors, the following people have contributed to this specification:

Recognize members of the Task Force? Not an easy to find list of contributors.

Changes between RDF 1.1 and RDF 1.2

A detailed overview of the differences between RDF versions 1.0 and 1.1 can be found in [[[RDF11-NEW]]] [[RDF11-NEW]].