A color-coded version of this document showing changes made since the previous version is also available.
This document is also available in these non-normative formats: PDF version.
Copyright © 2012 W3C® (MIT, ERCIM, Keio), All Rights Reserved. W3C liability, trademark and document use rules apply.
This document specifies a reversible mapping (or transformation) from Rule Interchange Format (RIF) XML documents to Resource Description Framework (RDF) graphs. This mapping allows the contents of RIF documents to be interoperably stored and processed as RDF triples, using existing serializations and tools for RDF. When used with the standard mapping from RDF triples to RIF frames, this also provides a "reflection" or "introspection" mechanism, an interoperable way for RIF rules to operate on RIF documents.
This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at http://www.w3.org/TR/.
This document is being published as one of a set of 12 documents:
There have been no changes to the body of this document since the previous version. For details on earlier changes, see the change log.
Please send any comments to public-rif-comments@w3.org (public archive). Although work on this document by the Rule Interchange Format (RIF) Working Group is complete, comments may be addressed in the errata or in future revisions. Open discussion among developers is welcome at public-rif-dev@w3.org (public archive).
Publication as a Working Group Note does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.
This document was produced by a group operating under the 5 February 2004 W3C Patent Policy. W3C maintains a public list of any patent disclosures made in connection with the deliverables of the group; that page also includes instructions for disclosing a patent. An individual who has actual knowledge of a patent which the individual believes contains Essential Claim(s) must disclose the information in accordance with section 6 of the W3C Patent Policy.
Table of Contents |
The Rule Interchange Format (RIF) [RIF Overview] is an interlingua between rule systems. It is an overlapping family of XML languages (called "dialects") designed for transmitting and storing various kinds of computer-processable rules and related data. Three standard dialects have been defined: RIF Core [RIF Core], RIF Basic Logic Dialect (BLD) [RIF BLD], and RIF Production Rules Dialect (PRD) [RIF PRD]. RIF Core is a sublanguage of BLD and of PRD: every Core document is also a BLD document and a PRD document.
RIF was envisioned [RIF Charter] to be extensible, allowing third parties to define non-standard extensions which could be combined into new dialects, as needed, to support interchange of rule sets which include features not defined in the standard dialects. Despite this vision, no general mechanism for extensions has been detailed in the RIF specifications. The RIF Framework for Logic Dialects (FLD) [RIF FLD] does, however, specify a way to create more expressive logic dialects.
The Resource Description Framework (RDF) [RDF] is a standard abstract way to represent data. The units of data in RDF are triples consisting of a subject, property (or predicate), and value (or object). These triples are similar to (and compatible with) RIF Frames (see RIF-RDF Combinations [RIF RDF+OWL]). A set of triples can be viewed as a directed labeled graph, where the nodes are subjects and values and the arcs are labeled with property identifiers; we therefore speak of a set of RDF triples as an RDF graph. RDF graphs can be serialized in multiple equivalent syntaxes, including RDF/XML [RDF XML], RDFa [RDFa], and Turtle [Turtle]. RDF can be processed with a wide variety of software tools [RDF Tools].
This specification defines a reversible mapping from RIF syntactic structures to RDF graphs. The definition is presented in tables where each row in the tables shows an XML template and a corresponding RDF graph template. The mapping is performed, roughly speaking, by finding the first matching XML template, then producing the corresponding graph. In some cases, the graph will require recursive translations of XML subtrees. The resulting graph has one node, called the focus node, which represents the XML root node, which in RIF Core, BLD, and PRD is rif:Document.
A reverse mapping, described in section 6, is possible for standard RIF by simply matching the RDF template and generating the corresponding XML. For extended RIF, so the reverse mapping can only done if the translator knows the XML grammar being generated.
Note that RDF serializations produced via this mapping are not standard RIF documents and cannot necessarily be understood by RIF implementations.
The rest of this document is structured as follows:
In designing this mapping a few use cases were considered:
The following requirements were taken into account in this design:
Conceptually, in RIF there are two kinds of extensions, divided by how consumers which do not implement the extension are to handle them. Extensions which non-implementing consumers may ignore are encoded in RIF metadata; extensions which non-implementing consumers must understand before processing require altering the syntax so that the RIF document will not be schema valid, such as by introducing new XML elements.
In the RDF mapping, this distinction must be made somewhat differently. In RDF, may-ignore extensions can be encoded with additional triples about RIF syntactic elements, while must-understand extensions require removing or replacing properties required for non-extended decoding. This absence of a required arc prevents the decoding to a schema-valid XML RIF document, causing non-implementing consumers to abort.
These restrictions are necessary in order to meet the stated requirements. In particular, without these restrictions, a RIF document (in RDF graph form) being transformed by an incomplete reasoner into another RIF document (also in RDF graph form) could produce an unintended and incorrect result just because the reasoning was incomplete. With these restrictions, the result will not match the reverse-mapping until it is sufficiently complete.
The mapping from RIF XML to RDF Graphs is expressed as a function Tr:
Tr(rif-xml-tree) → <focus-node, triples>
For every standard RIF Document, and for certain subtrees of RIF documents and extended RIF documents, Tr maps to the pair of an RDF node and an RDF graph. The node, called the focus node represents the same syntactic element as the root of the given XML tree. The RDF graph is a standard RDF graph, a set of RDF triples, and always contains the focus node. The focus node is usually a fresh blank node, but it might have a IRI label in certain cases, as detailed below.
In this document, the Turtle [Turtle] RDF serialization syntax is used for expressing triples and graphs. Turtle has a very terse syntax for lists, ( item-1 ... item-n ) and for fresh blank nodes and the triples using them as the subject: [ property-1 value-1; ... property-n value-n ]. These constructs allow the mapping to be presented and examples to be shown with relative simplicity.
The mapping is defined recursively, with each application of Tr converting an XML class element to a focus node, with additional triples. Class elements in RIF XML have tags that begin with an uppercase letter and represent a particular syntactic entity. Except for rif:Var and rif:Const class elements (detailed in Table 1, below), all the class elements follow a general form, containing a sequence of property elements, each containing additional class elements. The mapping for these is detailed in Table 2 and Table 3, below.
For another example of a specification of a mapping to RDF graphs, which may lend insight into how to use this specification, see OWL 2 Mapping to RDF Graphs [OWL2 Mapping].
If a system performing the mapping has determined the input document to be in one of the three standard dialects (Core, PRD, BLD), the rdf:type of the root focus-node may be set corresponding to that dialect: rif:CoreDocument, rif:PRDDocument, or rif:BLDDocument, instead of rif:Document. These classes for the standard dialects are defined to be disjoint: rif:BLDDocument contains those BLD documents which are not Core documents, and rif:PRDDocument contains those PRD documents which are not Core documents. Extensions may result in additional rdf:type arcs on the root focus-node.
Note that RIF XML allows for relative IRIs, which are expanded to absolute IRIs using the xml:base directive. This expansion must be done before Tr.
All standard elements in RIF XML have the namespace "http://www.w3.org/2007/rif#", and the attributes have no namespace. Extensions are expected to use other namespaces for the elements and are not allowed to introduce new attributes.
The RIF-in-RDF mapping produces RDF graphs that use the same namespace, although they use that namespace name in the normal RDF way (as an IRI prefix) instead of in the XML way (as a disambiguator).
By keeping the namespace the same, transformation software can correctly operate, without modification, on all RIF documents, even ones containing extensions.
Note that this use of the same namespace means that in certain cases RIF and RDF/XML documents cannot be distinguished simply by their namespace use. Moreover, since the rdf:RDF root element is optional in RDF/XML, in some cases it is not possible to distinguish between RIF and RDF/XML documents just by schema-validating or RDF-parsing the XML. In those cases, additional inspection of the structure is necessary. In general, systems should therefore be careful to maintain external file type information. This is typically done with either the media types ("application/rif+xml" and "application/rdf+xml") or the suggested filename extensions (".rif" and ".rdf").
In the tables below, the followng XML DOCTYPE declaration is assumed, allowing for abbreviation of the RDF and RIF namespaces:
<!DOCTYPE rif:Document [ <!ENTITY rdf "http://www.w3.org/1999/02/22-rdf-syntax-ns#"> <!ENTITY rif "http://www.w3.org/2007/rif#"> ]>
Also, the default XML namespace is assumed to be "http://www.w3.org/2007/rif#" and for use in Turtle, the following prefix declarations are assumed to be in effect:
@prefix rif: <http://www.w3.org/2007/rif#> @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> @prefix xs: <http://www.w3.org/2001/XMLSchema#>
Any RIF class element may have a first child of the form:
<id> <Const type="&rif;iri">id</Const> </id>
When this child is present, it is ignored for other processing and the id text is used as the IRI (URI-Reference) label for the focus_node, instead of it being left as a blank node.
As a special case, if the rif:Document element does not have an <id> child, its focus_node should be given the Web address (IRI) of the input XML document, if it has one.
After this optional <id> element, any RIF class element may contain a <meta> element. This element is processed according to general element processing rules, below.
Systems performing this transformation may also attempt to convert the metadata to RDF using the standard Frame-RDF correspondence [RIF RDF+OWL] , and include it in the returned triples. The conversion is not always possible, because some frame formulas are not expressible in RDF; systems which attempt this transformation and encounter such frame formulas in metadata should issue a warning. Even if the frames are converted to RDF like this, implementations must, under default settings, still keep the rif:meta triples intact, to support stable roundtripping.
Table 1, below, defines a portion of Tr. When a rif-xml-tree, X, matches the entry in column one, treating terms written like-this as metavariables, the result of Tr(X) is the pair <focus_node,G>, where G is the set of triples indicated by the second column.
Note that, although not shown in this table, <id> and <meta> child elements are allowed before the text in these elements. Additional elements after <id> and <meta> and before the character data may be allowed by extended schemas and must be processed as normal (extended) property elements.
Note that metadata about Consts is applied to the focus_node, not to the RDF literal produced. For example an explanation comment on a Const is understood to explain why that value is used in that spot, not state general properties of that value.
Input XML Pattern, X | Output RDF Triples, G |
---|---|
<Var>variable-name</Var> |
focus_node rdf:type rif:Var focus_node rif:varname "variable-name" |
<Const type="&rif;iri">value</Const> |
focus_node rdf:type rif:Const focus_node rif:constIRI "value" Note that the value is an absolute IRI. Relative IRIs must be converted using the xml:base. |
<Const type="&rif;local">value</Const> |
focus_node rdf:type rif:Const focus_node rif:constname "value" |
<Const type="&rdf;PlainLiteral">text@</Const> |
focus_node rdf:type rif:Const focus_node rif:value "value" |
<Const type="&rdf;PlainLiteral">text@langtag</Const> |
focus_node rdf:type rif:Const focus_node rif:value "value"@langtag |
<Const type="type-iri">value</Const> |
focus_node rdf:type rif:Const focus_node rif:value "value"^^type-iri |
Except as noted above, the Tr mapping for any class element (denoted as parent in Table 2) is to a new focus node and a set of triples which depend on each of the children of the class element. The exact dependency is detailed in this section.
Each child of the class element being mapped (except as noted above) is a property element (denoted as child in Table 2). There are four kinds of property elements:
The mapping for each mode is specified in Table 2 below. The mapping depends on the identity of an RDF property, written as prop, and the mode. Table 3 specifies special-case values for prop and mode, but otherwise they are determined as follows:
Mode | Property Element XML Pattern (With Parent Shown) | RDF Triples added to Tr result |
---|---|---|
0 |
. . . <parent> . . . <child ordered="yes"> item-1 . . . item-n </child> . . . </parent> . . . |
focus_node rdf:type parent focus_node prop ( id-1 . . . id-n ) triples-1 . . . triples-n where: <id-1, triples-1> = Tr(item-1) . . . <id-n, triples-n> = Tr(item-n) |
1 |
. . . <parent> . . . <child>item</child> . . . </parent> . . . |
focus_node rdf:type parent focus_node prop id triples where: <id, triples> = Tr(item) As a special case, if item is merely character data (with no XML elements), id is an RDF plain literal with the same value and triples-n is empty. This occurs in RIF Core, for example, with the rif:location property element. |
2 |
. . . <parent> . . . <child>item_1</child> . . . <child>item_n</child> . . . </parent> . . . |
focus_node rdf:type parent focus_node prop ( id-1 . . . id-n ) triples-1 . . . triples-n where: <id-1, triples-1> = Tr(item-1) ... <id-n, triples-n> = Tr(item-n) |
3 | As found in <Atom> and <Expr> in BLD (but not Core or PRD):
. . . <parent> . . . <slot ordered="yes"> <Name>name-1</Name> value-1 </slot> . . . <slot ordered="yes"> <Name>name-n</Name> value-n </slot> . . . <parent> . . . |
focus_node rdf:type parent focus_node rif:namedargs ( [ rdf:type rif:NamedArg; rif:argname "name-1"; rif:argvalue value-id-1 ] . . . [ rdf:type rif:NamedArg; rif:argname "name-n"; rif:argvalue value-id-n ] ) triples-1 . . . triples-n where: <value-id-1, triples-1> = Tr(value-1) . . . <value-id-n, triples-n> = Tr(value-n) |
3 |
... <Frame> . . . <slot ordered="yes"> key-1 value-1 </slot> . . . <slot ordered="yes"> key-n value-n </slot> . . . </Frame> ... |
focus_node rdf:type rif:Frame. focus_node rif:slots ( [ rdf:type rif:Slot; rif:slotkey nk-1; rif:slotvalue nv-1 ] . . . [ rdf:type rif:Slot; rif:slotkey nk-n; rif:slotvalue nv-n ] ) tk-1 . . . tk-n tv-1 . . . tv-n where: <nk-1, tk-1> = Tr(key-1) . . . <nk-n, tk-n> = Tr(key-n) <nv-1, tv-1> = Tr(value-1) . . . <nv-n, tv-n> = Tr(value-n) |
This table specifies exceptions to the default rules for determining the value of prop and the mode of the property element:
Class Element (parent) | Property Element (child) | RDF Property (prop) | Mode |
---|---|---|---|
Document | directive | rif:directives | 2 |
Group | sentence | rif:sentences | 2 |
Forall | declare | rif:vars | 2 |
Exists | declare | rif:vars | 2 |
And | formula | rif:formulas | 2 |
Or | formula | rif:formulas | 2 |
Frame | slot | rif:slots | 3 |
Atom | slot | rif:namedargs | 3 |
Expr | slot | rif:namedargs | 3 |
We call the inverse mapping XTr:
XTr( focus-node, triples ) → rif-xml-tree
For any RIF-XML document D, valid according to some RIF dialect XML schema, given the transformation:
D' = Xtr( Tr(D) )
D' and D will differ only in presentation aspects (non-significant whitespace, XML comments, IRIs being made absolute using xml:base, etc), unrelated to the semantics. That is, the semantics of D will be preserved so the rule sets corresponding to D and D' will have identical entailments under the relevant RIF Dialect semantics.
The ordering of child property elements by Xtr is determined by the relevant dialect schema, if known; otherwise they are lexicographic, sorted first by namespace then by local part. Xtr implementations may accept schema parameters to constrain their extraction to conform to a particular dialect's schema, and handle non-lexicographic ordering, like:
XTr( focus-node, triples, XML-schema, XML-root-element-in-schema) → rif-xml-tree
If they do not do this, they will not be able to correctly extract RIF XML documents in non-standard schemas using non-lexicographic element ordering. (For this reason, extensions are advised to use lexicographic ordering of elements in their schema.) For the three standard dialects, Xtr implementations must emit the children in schema-valid order.
RIF RDF and OWL Compatibility [RIF RDF+OWL] defines the entailments of combinations (R, G) where R (a RIF rule set) includes an import of G (an RDF graph).
We hereby define an RDF predicate rif:usedWithProfile
which enables an import to be specified from the graph G instead of from R. This definition also appears in SPARQL 1.1 Entailment Regimes [SPARQL ER].
In the simple usage the graph G is a plain RDF graph and rif:usedWithProfile
is used to combine that graph with one or more externally defined RIF rule sets. In this usage each subject of a rif:usedWithProfile
assertion should be the URI for a RIF rule set (which may be encoded in RIF-XML or RIF-in-RDF) and the object should be an import profile as defined in RIF RDF and OWL Compatibility [RIF RDF+OWL].
It is also possible for the graph G to itself contain an encoded ruleset along with additional RDF statements to which the ruleset is intended to apply. If graph G is obtained from a base IRI Ug
then the statement:
Ug rif:usedWithProfile P .
within the graph causes it to be treated as both the graph G and the source of the rule set R in the combination (R, G). Syntactically such a statement can be made by using the empty relative IRI reference <>
provided that the document base IRI has been set appropriately.
The semantics of rif:usedWithProfile
is explained in the following section.
Note: this definition also appears in SPARQL 1.1 Entailment Regimes [SPARQL ER].
A RIF-in-RDF-aware processor shall treat any RDF graph G as a RIF-RDF or RIF-OWL combination (cf. [RIF RDF+OWL]) as follows:
Let G' be the graph obtained from G by removing all triples with predicate rif:usedWithProfile
.
Then G is to be treated by a RIF-in-RDF-aware processor as the ruleset R:
Document ( Imports(R1') ... Imports(Rn') Imports(G' P1) ... Imports(G' Pn) )
Where Ri and Pi are the subjects/objects respectively of triples of form:
Ri rif:usedWithProfile Pi
and Ri' denotes
Remark 1: Note that the fact that G' is treated as being imported with all profiles P1 ... Pn enforces G' to be treated according to the highest profiles among P1 ... Pn, cf. Section 5.2 of [RIF RDF+OWL].
Remark 2:
If G also includes a rif:usedWithProfile
statement referring to itself (i.e. with subject
Ug
where the graph G can be obtained from Ug
) then the rules encoded in that document will be included in the rule imports but the encoding of the rules will remain visible within G'.
Remark 3: Note the discussion in the section 6 that the inversion of Tr is not a deterministic function.
Remark 4: Note that in the case where the graph G includes encoded RIF rules then, as a result of RDF graph merge, it may encode more than one RIF document. A RIF-in-RDF process MAY choose to combine all the rules in each documents into a single RIF document or MAY issue a warning. Note that future RIF dialects may have semantics which depend on rule ordering.
This document is the product of the Rules Interchange Format (RIF) Working Group (see below) whose members deserve recognition for their time and commitment. The editor extends special thanks to Dave Reynolds for his particularly attentive and insightful review comments.
The regular attendees at meetings of the Rule Interchange Format (RIF) Working Group at the time this document was developed were: Adrian Paschke (Freie Universitaet Berlin), Axel Polleres (DERI), Chris Welty (IBM), Christian de Sainte Marie (IBM), Dave Reynolds (HP), Gary Hallmark (ORACLE), Harold Boley (NRC), Jos de Bruijn (FUB), Leora Morgenstern (IBM), Michael Kifer (Stony Brook), Mike Dean (BBN), Sandro Hawke (W3C/MIT), and Stella Mitchell (IBM).
[
Examples, test cases, and links to implementations may be found at http://www.w3.org/2011/rif-in-rdf.
Here is Example 8 from BLD converted via the RIF-in-RDF mapping to Turtle:
@prefix : <http://www.w3.org/2007/rif#> . @prefix xs: <http://www.w3.org/2001/XMLSchema#> . [] a :Document; :directives (); :payload <http://sample.org>. <http://sample.org> a :Group; :meta [ a :Frame; :object [ a :Const; :constname "pd" ]; :slots ( [ a :Slot; :slotkey [ a :Const; :constIRI "http://purl.org/dc/terms/publisher"^^xs:anyURI ]; :slotvalue [ a :Const; :constIRI "http://www.w3.org/"^^xs:anyURI ] ] [ a :Slot; :slotkey [ a :Const; :constIRI "http://purl.org/dc/terms/date"^^xs:anyURI ]; :slotvalue [ a :Const; :value "2008-04-04"^^xs:date ] ] ) ]; :sentences ( [ a :Forall; :formula [ a :Implies; :if [ a :And; :formulas ( [ a :Atom; :args ( [ a :Var; :varname "item" ] ); :op [ a :Const; :constIRI "http://example.com/concepts#perishable"^^xs:anyURI ] ] [ a :Atom; :args ( [ a :Var; :varname "item" ] [ a :Var; :varname "deliverydate" ] [ a :Const; :constIRI "http://example.com/John"^^xs:anyURI ] ); :op [ a :Const; :constIRI "http://example.com/concepts#delivered"^^xs:anyURI ] ] [ a :Atom; :args ( [ a :Var; :varname "item" ] [ a :Var; :varname "scheduledate" ] ); :op [ a :Const; :constIRI "http://example.com/concepts#scheduled"^^xs:anyURI ] ] [ a :Equal; :left [ a :Var; :varname "diffduration" ]; :right [ a :External; :content [ a :Expr; :args ( [ a :Var; :varname "deliverydate" ] [ a :Var; :varname "scheduledate" ] ); :op [ a :Const; :constIRI "http://www.w3.org/2007/rif-builtin-function#subtract-dateTimes"^^xs:anyURI ] ] ] ] [ a :Equal; :left [ a :Var; :varname "diffdays" ]; :right [ a :External; :content [ a :Expr; :args ( [ a :Var; :varname "diffduration" ] ); :op [ a :Const; :constIRI "http://www.w3.org/2007/rif-builtin-function#days-from-duration"^^xs:anyURI ] ] ] ] [ a :External; :content [ a :Atom; :args ( [ a :Var; :varname "diffdays" ] [ a :Const; :value 10 ] ); :op [ a :Const; :constIRI "http://www.w3.org/2007/rif-builtin-predicate#numeric-greater-than"^^xs:anyURI ] ] ] ) ]; :then [ a :Atom; :args ( [ a :Const; :constIRI "http://example.com/John"^^xs:anyURI ] [ a :Var; :varname "item" ] ); :op [ a :Const; :constIRI "http://example.com/concepts#reject"^^xs:anyURI ] ] ]; :vars ( [ a :Var; :varname "item" ] [ a :Var; :varname "deliverydate" ] [ a :Var; :varname "scheduledate" ] [ a :Var; :varname "diffduration" ] [ a :Var; :varname "diffdays" ] ) ] [ a :Forall; :formula [ a :Implies; :if [ a :Atom; :args ( [ a :Var; :varname "item" ] ); :op [ a :Const; :constIRI "http://example.com/concepts#unsolicited"^^xs:anyURI ] ]; :then [ a :Atom; :args ( [ a :Const; :constIRI "http://example.com/Fred"^^xs:anyURI ] [ a :Var; :varname "item" ] ); :op [ a :Const; :constIRI "http://example.com/concepts#reject"^^xs:anyURI ] ] ]; :vars ( [ a :Var; :varname "item" ] ) ] ) .