Schema Specification of Deliberation RuleML Version 1.0

An introduction to RuleML is given in our &primer;. Also, the paper &overarching; discusses the upper layer of the RuleML hierarchy of rules. In that terminology, the system of RuleML languages presented here covers derivation rules, which includes Horn Logic, and part of deliberation rules, in particular, First-Order Logic (FOL) with equality and negation-as-failure.

This is because we think it is important to start with a subset of simple rules (derivation rules), test and refine our principal strategy using these, and then proceed to more general categories of rules in the hierarchy (deliberation rules), as well as to other kinds of rules (reaction rules).

A comprehensive &glossary; accompanies this specification.

The grammar of Version 1.0 is partially described in the &content-models; document, where the content model of each individual element is compared across the named sublanguages, as specified within the &xsd-schemas;. The RuleML grammar is also partially described by the &relaxng-schemas;.

Below is a summary of the changes in Version 1.0:

Readers who already know Version 0.91 may want to refer to the Changes section.

Numerous sample RuleML documents have been prepared and maintained; some exemplify features of RuleML and are useful didactically while others are mostly for testing puposes. Updated examples (e.g., own.ruleml and reify.ruleml) accompany the Version 1.0 release and can be found with all the others in the Examples directory. The following examples are new:

Examples from previous versions of RuleML are also maintained, e.g. 0.91 examples).

Continuing the practice begun in 0.86, an XSLT stylesheet named 091-to-100.xslt has been created to automatically update Version 0.91 documents to Version 1.0. For instance, taking the discount example from 0.91 as input, the stylesheet outputs the following: discount_100.ruleml. Additional translation output samples are found in the Upgrader directory.

Upgrader transformation can be accomplished using a web-based XSLT tool, Online XSLT 2.0 Service, provided by W3C. Instructions for this process can be found in Appendix 4.

The stylesheet has also been tested using oXygen version 12.2, whereby it was confirmed that all examples in the directory http://ruleml.org/1.0/exa are properly upgraded. However, the upgraded instances documents will have the RuleML namespace as the default namespace, independent of the choice of prefix for the RuleML namespace in the original instance. A similar 091-to-ruleml100.xslt has been developed that uses the prefix "ruleml" for the RuleML namespace.

An XSLT processor which may be used to perform these transformations on a whole directory at once is Saxon, using the following command:

A use example of the Upgrader is described in the WineOnto migration path documentation.

RuleML has always allowed abbreviated encoding (skipped edge tags and default attribute values) and some freedom in the ordering of elements. The new Relax NG schemas allows even more freedom in element ordering than the XSD schemas, with all perturbations possible as long as the missing edge tags may be unambiguously reconstructed. An XSLT stylesheet has been developed (see RON, also see the &normalizer; directory) for normalizing the syntax used in a given Version 1.0 instance, undoing the abbreviated encoding and re-ordering of elements. An example of the normal form is the expanded version of the compact 'own' example.

The goals of this normalizer include the following:

We say that using this normalizer followed by schema-based RuleML validation performs "normalidation" on RuleML instances. That is to say, not only are missing edge tags and attributes with default values inserted into the Version 1.0 instance in order to normalize it, but the general tree structure of the file is also validated syntactically to ensure that Node and edge elements only appear in correct positions.

For example, the compact version of the 'own' example:

  
    
      
        
          buy
          person
          merchant
          object
        
        
          person
          keep
          object
        
      
      
        person
        own
        object
      
    
    ...
    
      Mary
      keep
      XMLBible
    
  

  ]]>

should be normalized to the expanded version:

  
  
      
        
          
            
              
                
                  buy
                  person
                  merchant	
                  object
                
              
              
                
                  keep
                  person
                  object
                
              
            
          
          
            
              own
              person
              object
            
          
        
      
      ...
      
        
          keep
          Mary
          XMLBible
        
      
  
  
]]>

XSLT processing will be accomplished using procedures identical to those of the XSLT Upgrader. Instructions for the W3C web-based process can be found in Appendix 4.

The &glossary; for Version 1.0 contains descriptions for every element in Version 1.0.

As in previous versions of RuleML, there are several pairs of attributes in Version 1.0 that have names which differ only by a 'map' prefix: @closure and @mapClosure, @direction and @mapDirection, and @material and @mapMaterial. The naming convention reflects attribute 'mapping' through XML trees, where

. . .. . .. . .]]>

basically acts like

. . .. . .. . .]]>

or

. . .. . .. . .]]>

The XSD schemas use a tree-based modularization strategy. The most current model of the XSD modularization of RuleML is always documented at http://ruleml.org/modularization/#Model.

The Relax NG schemas use a different modularization approach based on lattices. The details are given in the publication about the re-engineering of RuleML in Relax NG and the RuleML MYNG Wiki page. &myng; is accessible through a GUI or, for advanced users, a PHP script .

The &xsd-schemas; specification of Version 1.0 has been created using an approach that is consistent with that of all earlier XSDs, and back to that of the (version 0.85) DTDs. The &content-models; document should be read in parallel to the XSD schemas, because they give a high-level, complete documentation of the XSDs. Likewise, the &glossary; can help to find quick descriptions of, and cross-reference between, the Version 1.0 XML elements.

Appended below is a simple example rulebase (Appendix 1), the XSD Schema for the Datalog sublanguage of RuleML to which the rulebase conforms (Appendix 2), instructions for how to validate instances against the XSD schema (Appendix 3), instructions for how to transform with XSLT stylesheets (Appendix 4), and instructions for how to validate instances against the Relax NG schema (Appendix 5).

Source: own-xsd.ruleml

  

    

    
      
        
        
          
            buy
            person
            merchant
            object
          
          
            keep
            person
            object
          
        
      
      
        
          own
          person
          object
        
      
    

  

    
      
        
          sell
          merchant
          person
          object
        
      
      
        
          buy
          person
          merchant
          object
        
      
    
 
  
 

    
      sell
      John
      Mary
      XMLBible
    

 
  
 
    
      keep
      Mary
      XMLBible
    
  
 
]]>

Source: datalog.xsd

	
		
			XML Schema for the RuleML sublanguage "datalog".
			File: datalog.xsd
			Version: 1.0
			Last Modification: 2010-07-16
		
	

	

	

	
	
	
	
	
	
	
	

]]>

You should get the following output (using the default output):

Schema validating with XSV 3.1-1 of 2007/12/11 16:20:05

Schema resources involved

Attempt to load a schema document from http://ruleml.org/1.0/xsd/datalog.xsd (source: schemaLoc) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/performative_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/desc_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/quantifier_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/connective_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/atom_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/slot_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/term_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Attempt to load a schema document from http://ruleml.org/1.0/xsd/modules/iri_module.xsd (source: include) for http://ruleml.org/spec, succeeded

Changes in the Version 1.0 XSD release relative to the previous XSD version 0.91 are detailed below, including examples where appropriate.

The terminology "Type" and "role" (when referring to XML tags or elements) has been changed to "Node" and "edge", in order to avoid the confusion caused by simultaneous usage of multiple meanings of these words, as well as to emphasize the connection to graphs, especially the RDF graph model.

Schematron dependency has been reduced (as comments in the code). For example, the following code provides annotation related to restrictions on the interpretation attribute of nested functions.

  
    
      
        
          Functions nested within an uninterpreted function must also be uninterpreted.
        
      
    
  
]]>

Edge elements <head> and <body> are replaced with <then> and <if>, respectively. These elements appear as children of <Implies> and <Entails>. For example, the following implication:

  
    
      
        premium
        cust
      
      
        regular
        prod
      
    
  
  
    
      discount
      cust
      prod
      5.0 percent
    
  
]]>

expresses the statement "A premium customer recieves a 5% discount on regular products."

Edge elements <lhs> and <rhs> are replaced with <left> and <right>, respectively. These elements appear as children of <Equal>. As an example consider the statement:

    
  
    Lady Gaga
  
  
    Stefani Joanne Angelina Germanotta
  
]]>

which expresses that the individuals named "Lady Gaga" and "Stefani Joanne Angelina Germanotta" are the same.

In the SWSL languages, which are implemented only in the XSDs, Node elements <Hterm> and <Con> are replaced with <Uniterm> and <Const> respectively.

Attribute @in="no|semi|yes" with respective values are replaced with @per="copy|open|value". For example, the following functional expression uses the default value ("copy") of the attribute "per"

  MultiplyFn
  3
  X
]]>

to generate an algebraic expression that may be evaluated after the variable X is bound to a numerical value. In contrast, the expression

  MultiplyFn
  3
  4
]]>

will be evaluated (to <Data xsi:type="xs:integer">12</Data>) before inference is performed.

Attribute @uri is replaced with @iri. This attribute is used to identify, by Internationalized Resource Identifier (IRI), the elements <Ind>, <Rel> and <Fun> as individuals, relations and functions, respectively. For example:

  
    
  
  
  Ora Lassila
]]>

expresses a fact equivalent to the RDF triple:

  
    Ora Lassila
  
]]>

A new edge element <act>, with a required @index attribute, is introduced as an optional wrapper for all performatives of the <RuleML> element, as shown here:

  
    
  
]]>

A new attribute @node is introduced as an optional IRI identifier of a Node element, useful for referencing the element in external meta-knowledge. It is allowed on all Nodes, including leaf Nodes such as <Data>, <Reify> and <Rel>.

  ......
]]>

A new edge element <meta> is introduced as a non-skippable wrapper of a formula expressing meta-knowledge, allowed zero to many times in the header of non-leaf Node elements. When a slotted <Atom> is wrapped in <meta>, the object of the slotted frame is assumed (unless specified otherwise) to be the reification of the parent Node, which may be specified with the @node attribute, or may be anonymous.

  
    
      
      
      
      
  
  ......
]]>

The edge element <oid> is no longer allowed on all non-leaf Node elements, but is restricted to <Atom> and <Expr> (and <Uniterm> in the SWSL branch,) for use as an identifier for the object of slotted atoms and expressions. The functionality of providing an identifier for attaching meta-knowledge, such as the defeasible priority of rules, is transferred to the @node attribute.

The attributes @xml:base and @xml:id are allowed on all elements. The base for resolving relative IRI references is explicitly declared with @xml:base. The attribute @xml:id provides an alternate means of identifying elements, and is applicable to edges as well as nodes. The semantics of the @xml:id identifier, which denotes the information resource fragment that is the element and is referentially opaque, differs from that of the @node identifier, which denotes a reification of the logical or extra-logical entity associated with the element and is referentially transparent.

  
    
  
]]>

Additional freedom in element order with partial stripe-skipping has been introduced in the "relaxed form serialization", as specified in the Relax NG schemas. In particular, infix and postfix operator notation is now allowed in the relaxed form serialization as long as the <op> element is not skipped.

Features that are "deprecation candidates", i.e. that may be deprecated in future releases, include: <Reify> at Datalog and lower, as this introduces the possibility of nesting reification to an arbitrary level ; the default values of attributes which indicate semantics variants: @direction, @mapDirection, @material, @mapMaterial, @oriented, @val, @per, as it may be more effective to use the absence of the attribute to indicate a "neutral" semantics; the (SWSL) syntax, whose current syntax may be deprecated in favor of better integration with the rest of the RuleML syntax.

During the review of the Version 0.91 XSD schemas that was conducted in conjunction with the re-engineering of RuleML in Relax NG, a number of shortcomings and defects were identified. A number of these issues were patched in the Version 0.91 Patched release, and these patched XSDs form the basis of the hand-written XSDs for Version 1.0. There are several issues that could not be resolved in the handwritten XSDs, including: Reification in Languages Below Datalog: the Binary sublanguages have an overly-permissive schema for reification (<Reify>). For example, in bindatagroundfact, it is valid to reify a universal quantification (<Forall> ), while it is not allowed to construct a universal quantification in the language. It is not possibly to fix this issue without discarding the tree-based modularization approach of the hand-written schemas. Because Version 1.0 is a "Rosetta" release, we have retained the hand-written XSD schemas that contain this issue for consistency with legacy schemas. However, the Relax NG schemas are not affected by this issue, and thus there is a discrepancy between the XSD and Relax NG schemas in these unusual cases. Circular Group Definitions: all schemas that redefine the Datalog XSD schema use circular group definitions, leading to an inability to validate against these schemas using some validators, such as Xerces. Similar to the issue above, it is not possible to fix this issue within the modularization approach of the hand-written XSD schemas. However, XSV and Saxon validators are lax in enforcing Constraints on Model Group Schema Components (§3.8.6) of the XSD specification, Part 1, and are able to work around it. Relax NG schemas and the XSD schemas automatically generated from the Relax NG schemas are not affected by this issue. The latter two issues will be resolved in Version 1.1 by discontinuing support of the handwritten XSDs, relying instead on XSD schemas automatically generated from the Relax NG schemas.