This section is non-normative.
This section introduces the RDF/XML syntax, describes how it
encodes RDF graphs and explains this with examples. If there is any
conflict between this informal description and the formal description
of the syntax and grammar in sections
5. Syntax Data Model and
6. RDF/XML Grammar,
the latter two sections take precedence.
This document specifies the syntax of RDF/XML as a grammar on an
alphabet of symbols. The symbols are called events in the
style of the XPATH
Information Set Mapping.
A sequence of events is normally derived from an XML document, in
which case they are in document order as defined below in
5.2 Information Set Mapping.
The sequence these events form are intended to be similar to the sequence
of events produced by the [SAX] XML API from
the same XML document. Sequences of events may be checked against
the grammar to determine whether they are or are not syntactically
well-formed RDF/XML.
The RDF 1.2 Concepts and Abstract Syntax specification [RDF12-CONCEPTS] defines four types of RDF terms:
IRIs,
literals,
blank nodes, and
triple terms.
Literals are composed of a lexical form
and an optional language tag [BCP47]
– possibly including a base direction –
or datatype IRI.
An extra type, prefix, is used during parsing to map string identifiers to namespace IRIs.
The grammar productions may include actions which fire when the
production is recognized. Taken together these actions define a
transformation from any syntactically well-formed RDF/XML sequence of
events into a set of triples by mapping matching productions
to RDF terms or their components (e.g., language tags, lexical forms of literals).
Grammar productions change the parser state and emit triples.
The model given here illustrates one way to create a representation of
an RDF Graph
from an RDF/XML document. It does not mandate any implementation
method — any other method that results in a representation of the same
RDF Graph may be used.
In particular:
- This specification permits any
representation of an RDF graph;
in particular, it does not require the use of N-Triples [RDF12-N-TRIPLES].
- This specification does not require the use of
[XPATH] or [SAX]
- This specification places no constraints on the order in which
software transforming RDF/XML into a representation of a graph,
constructs the representation of the graph.
- Software transforming RDF/XML into a representation of a graph
MAY eliminate duplicate predicate arcs.
The syntax does not support non-well-formed XML documents, nor
documents that otherwise do not have an XML Information Set [XML-INFOSET]; for
example, that do not conform to
Namespaces in XML [XML-NAMES].
The Infoset requires support for
XML Base [XMLBASE].
RDF/XML uses the information item property [base IRI], discussed in
4.3 Resolving IRIs
This specification requires an
XML Information Set [XML-INFOSET]
which supports at least the following information items and
properties for RDF/XML:
- document information item
- [document element], [children], [base IRI]
- element information item
- [local name], [namespace name], [children], [attributes], [parent], [base IRI]
- attribute information item
- [local name], [namespace name], [normalized value]
- character information item
- [character code]
There is no mapping of the following items to data model events:
Other information items and properties have no mapping to
syntax data model events.
Element information items with reserved XML Names
(see Name in Extensible Markup Language (XML) 1.1 (Second Edition))
are not mapped to data model element events.
These are all those with property [prefix] beginning with xml
(case independent comparison) and all those with [prefix] property
having no value and which have [local name] beginning with
xml (case independent comparison).
All information items contained inside XML elements matching the
parseTypeLiteralPropertyElt
production form XML literals and do not follow this mapping.
See parseTypeLiteralPropertyElt
for further information.
This section is intended to satisfy the requirements for
Conformance
in the [XML-INFOSET] specification.
It specifies the information items and properties that are needed
to implement this specification.
There are nine types of event defined in the following subsections.
Most events are constructed from an Infoset information item (except
for IRI,
blank node,
literal with no explicit datatype and
typed literal). The effect
of an event constructor is to create a new event with a unique identity,
distinct from all other events. Events have accessor operations on them
and most have the string-value, IRI, rdf-term, literal, or bnode accessors that may be a static value
or computed.
Constructed from a
document information item
and takes the following accessors and values.
- document-element
- Set to the value of document information item property [document-element].
- children
- Set to the value of document information item property [children].
- base-iri
- Set to the value of document information item property [base IRI].
- language
- Set to the empty string.
- direction
- Set to the empty string.
- graph
- The graph used for collecting emitted triples,
initialized to a new RDF graph.
Constructed from an
element information item
and takes the following accessors and values:
- local-name
- Set to the value of element information item property [local name].
- namespace-name
- Set to the value of element information item property [namespace name].
- children
- Set to the value of element information item property [children].
- parent
- Set to the value of element information item property [parent].
- base-iri
- Set to the value of element information item property [base IRI].
- attributes
-
Made from the value of element information item
property [attributes] which is a set of attribute
information items.
If this set contains an attribute information item
xml:lang ([namespace name] property with the value
"http://www.w3.org/XML/1998/namespace" and
[local name] property value "lang")
it is removed from the set of attribute information items and the
·language·
accessor is set to the [normalized-value] property of the
attribute information item.
If the element has a
version accessor
with the value "1.2-basic", "1.2" or greater,
this set contains an attribute information item
its:dir ([namespace name] property with the value
"http://www.w3.org/2005/11/its" and
[local name] property value "dir")
it is removed from the set of attribute information items and the
·direction·
accessor is set to the [normalized-value] property of the
attribute information item,
which MUST be one of "ltr", "rtl", or the empty string.
If the element has a
version accessor
with the value "1.2-basic", "1.2" or greater,
this set contains an attribute information item
its:version ([namespace name] property with the value
"http://www.w3.org/2005/11/its" and
[local name] property value "version")
it is removed from the set of attribute information items and the
·its-version·
accessor is set to the [normalized-value] property of the
attribute information item.
All remaining reserved XML Names
(see Name in Extensible Markup Language (XML) 1.1 (Second Edition))
are now removed from the set.
These are, all attribute information items in the set with property [prefix]
beginning with xml
(case independent comparison) and all attribute information items with [prefix]
property having no value and which have [local name] beginning with
xml (case independent comparison) are removed.
Note that the [base IRI] accessor is computed by XML Base before any
xml:base attribute information item is deleted.
The remaining set of attribute information items are then used
to construct a new set of
Attribute Events
which is assigned as the value of this accessor.
- IRI
- Set to an IRI constructed from the concatenation of the
value of the namespace-name accessor and the value of the
local-name accessor.
- li-counter
- Set to the integer value 1.
- language
- Set from the
·attributes·
as described above.
If no value is given from the attributes, the value is set to the value of
the language accessor on the parent event (either a
Root Event or an
Element Event), which may be the empty string.
- direction
- Set from the
·attributes·
as described above.
If no value is given from the attributes, the value is set to the value of
the direction accessor on the parent event (either a
Root Event or an
Element Event), which may be the empty string.
- version
- Set from the
·attributes·
as described above.
If no value is given from the attributes, the value is set to the value of
the version accessor on the parent event (either a
Root Event or an
Element Event).
- its-version
- Set from the
·attributes·
as described above.
If no value is given from the attributes, the value is set to the value of
the version accessor on the parent event (either a
Root Event or an
Element Event), which SHOULD be set to
"2.0".
- subject
- Has no initial value. Takes a value that is an
Identifier event,
or a triple term.
This accessor is used on elements that deal with one node in the RDF graph,
this generally being the subject of an
RDF triple.
- graph
- The graph used for collecting emitted triples.
The initial value is taken from the parent event
(either a Root Event or an
Element Event),
or a new RDF graph, if not set.
May be modified by a
parseType="Triple" attribute.
Has no accessors. Marks the end of the containing element in
the sequence.
Constructed from an
attribute information item
and takes the following accessors and values:
- local-name
- Set to the value of attribute information item property [local name].
- namespace-name
- Set to the value of attribute information item property [namespace name].
- string-value
- Set to the value of the attribute information item
property [normalized value] as specified by [XML11] (if an attribute whose normalized
value is a zero-length string, then the string-value is also
a zero-length string).
- IRI
-
If ·namespace-name· is present,
set to an IRI constructed from the concatenation of the value of the
·namespace-name· accessor
and the value of the
·local-name· accessor.
Otherwise if ·local-name· is
ID, about,
parseType, resource, type,
or version,
set to an IRI constructed from the
value of the concatenation of the
RDF namespace IRI
and the value of the local-name accessor.
Other non-namespaced
·local-name· accessor values are
forbidden.
The support for a limited set of non-namespaced names is
REQUIRED and intended to allow RDF/XML documents specified in
[RDF-SYNTAX-GRAMMAR-19990222] to remain valid; new documents
SHOULD NOT use these unqualified attributes and applications MAY
choose to warn when the unqualified form is seen in a document.
The construction of IRIs from XML attributes can generate the same
IRIs from different XML attributes. This can cause ambiguity in the
grammar when matching attribute events (such as when
rdf:about and about XML attributes are
both present). Documents that have this are illegal.
- rdf-term
-
The value is an IRI
constructed from
string-value
resolved against the ·base-iri·
accessor of the containing element.
Constructed from a sequence of one or more consecutive
character information items.
Has the single accessor:
- string-value
- Set to the value of the string made from concatenating the
[character code]
property of each of the character information items.
An event for an IRI which has the following accessors:
- identifier
- Takes a string value used as an IRI.
- rdf-term
- The value is an IRI constructed from the
value of the ·identifier· accessor.
These events are constructed by giving a value for the
·identifier· accessor.
For further information on identifiers in RDF, see
4.2 Identifiers.
An event for an triple term which has the following accessors:
- rdf-term
- The value is a triple term.
These events are constructed by giving a value to the rdf-term accessor.
An event for a blank node identifier
which has the following accessors:
- identifier
- Takes a string value.
- blank-node
- The value is a blank node with local identifier taken from the
·identifier· accessor.
Within an RDF/XML document,
all blank nodes identified by the same identifier represent the same blank node.
These events are constructed by giving a value for the
·identifier· accessor.
For further information on identifiers in RDF, see
4.2 Identifiers.
An event for a typed literal which can have the following accessors:
- literal-value
- Takes a string value.
- literal-datatype
- Takes a string value used as an IRI.
- literal
-
The value is an RDF literal with the
lexical form taken from the value of the
·literal-value· accessor
and datatype IRI
taken from the value of the ·literal-datatype· accessor.
These events are constructed by giving values for the
·literal-value·
and ·literal-datatype· accessors.
Note
Interoperability Note (Informative):
Literals beginning with a Unicode combining character are
allowed however they may cause interoperability problems.
See [CHARMOD] for further information.
Note
Implementation Note (Informative):
In XML Schema (part 1) [XMLSCHEMA11-2],
white space normalization
occurs during validation according to the value of the whiteSpace
facet. The syntax mapping used in this document occurs after this,
so the whiteSpace facet formally has no further effect.
To transform the Infoset into the sequence of events
in document order, each
information item is transformed as described above to generate a
tree of events with accessors and values. Each element event is
then replaced as described below to turn the tree of events
into a sequence in document order.
- The original element event
- The value of the
children
accessor recursively transformed, a possibly empty ordered list of events.
- An end element event
The following notation is used to describe matching the sequence
of data model events as given in 5. Syntax Data Model
and the actions to perform for the matches.
The RDF/XML grammar is defined in terms of mapping from these matched
data model events to triples, using notation of the form:
number event-type event-content
where the event-content is an expression matching
event-types (as defined in 5.1 Events),
using notation given in the following sections.
The number is used for reference purposes.
The grammar action may include generating
new triples to the graph.
The following sections describe the general notation used and that
for event matching and actions.
| Notation |
Meaning |
| A == B |
Event accessor A matches expression B. |
| A != B |
A is not equal to B. |
| A | B | ... |
The A, B, ... terms are alternatives. |
| A - B |
The terms in A excluding all the terms in B. |
| anyURI. |
Any URI. |
| anyString. |
Any string. |
| list(item1, item2, ...); list() |
An ordered list of events. An empty list. |
| set(item1, item2, ...); set() |
An unordered set of events. An empty set. |
| * |
Zero or more of preceding term. |
| ? |
Zero or one of preceding term. |
| + |
One or more of preceding term. |
root(acc1 == value1,
acc2 == value2, ...) |
Match a Root Event with accessors.
|
start-element(acc1 == value1,
acc2 == value2, ...)
children
end-element() |
Match a sequence of
Element Event with accessors,
a possibly empty list of events as element content and an
End Element Event.
|
attribute(acc1 == value1,
acc2 == value2, ...) |
Match an Attribute Event
with accessors. |
| text() |
Match a Text Event. |
| Notation |
Meaning |
| A := B |
Assigns A the value B. |
| concat(A, B, ..) |
A string created by concatenating the terms in order. |
| resolve(e, s) |
An IRI created by interpreting s as a relative IRI reference to the
·base-iri· accessor of e
as defined in 4.3 Resolving IRIs. |
| generated-blank-node-id() |
A string value for a new distinct generated
blank node identifier
as defined in 4.2 Identifiers.
|
| event.accessor := value |
Sets an event accessor to the given value. |
| iri(identifier := value) |
Create a new IRI Event. |
| bnodeid(identifier := value) |
Create a new Blank Node Identifier Event.
See also section 4.2 Identifiers. |
literal(literal-value := string,
literal-language := language,
literal-direction := direction, ...) |
Create a new Literal Without Datatype Event. |
| string(iri) |
The string value of an IRI. |
| typed-literal(literal-value := string, ...) |
Create a new Typed Literal Event. |
| triple(s, p, o) |
Create a new Triple Term Event composed of s, p, and o. |
This section is non-normative.
This section is non-normative.
This document is a product of extended deliberations by the RDF
Core working group, whose members have included: Art Barstow (W3C)
Dave Beckett (ILRT), Dan Brickley (W3C/ILRT), Dan Connolly (W3C),
Jeremy Carroll (Hewlett Packard), Ron Daniel (Interwoven Inc), Bill
dehOra (InterX), Jos De Roo (AGFA), Jan Grant (ILRT), Graham Klyne
(Clearswift and 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), 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), RV 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), 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.), Lauren Wood (SoftQuad).
This section is non-normative.
In addition to the editors, the following people have contributed to this specification:
Denis Ah-Kang, Dominik Tomaszuk, Peter F. Patel-Schneider, Pierre-Antoine Champin, and Ted Thibodeau Jr
Members of the RDF & SPARQL Working Group Group included
Vladimir Alexiev,
James Anderson,
Amin Anjomshoaa,
Julián Arenas-Guerrero,
Dörthe Arndt,
Bilal Ben Mahria,
Erich Bremer,
Dan Brickley,
Kurt Cagle,
Sarven Capadisli,
Rémi Ceres,
Pierre-Antoine Champin,
David Chaves-Fraga,
Souripriya Das,
Daniil Dobriy,
Enrico Franconi,
Jeffrey Phillips Freeman,
Fabien Gandon,
Benjamin Goering,
Damien Graux,
Adrian Gschwend,
Olaf Hartig,
Timothée Haudebourg,
Ian Horrocks,
Gregg Kellogg,
Mark Kim,
Jose Emilio Labra Gayo,
Ora Lassila,
Richard Lea,
Niklas Lindström,
Pasquale Lisena,
Thomas Lörtsch,
Matthew Nguyen,
Peter Patel-Schneider,
Thomas Pellissier Tanon,
Dave Raggett,
Jean-Yves ROSSI,
Felix Sasaki,
Andy Seaborne,
Alan Snyder,
Stuart Sutton,
Ruben Taelman,
Ted Thibodeau Jr,
Dominik Tomaszuk,
Raphaël Troncy,
William Van Woensel,
Gregory Williams,
Jesse Wright,
Achille Zappa, and
Antoine Zimmermann.
Editor's note
Recognize members of the Task Force? Not an easy to find list of contributors.
