Jingzhi Guo's Homepage teaching research publication demo
CONEX Resource

Syntax of XML PRODUCT MAP [2]

Jingzhi Guo

As a specific implementation of generic CONEX grammar, XML PRODUCT MAP (XPM) fol-lows DTD syntax that conforms to W3C XML recommendation [1]. An XPM specification represents a concept tree, which includes three major types of XML nodes: element node, attribute node and text node. In XPM, the local name of an element node represents the name of a concept structure, and the attribute list of the element node represents the n-tuple of denotative concept structure (i.e. representing a meta-concept structure). The text node of an element node represents a concept value of a reified concept. The XML tree structure repre-sents the denotation-connotation relationship, which is structurally abstracted as concept classifier structure. In the following, the syntax of XML PRODUCT MAP is described.

(1) Tokenised Meta-Structure: A tokenised meta-structure of XPM is a token of the decla-ration of meta-structure corresponding to a structure declaration StructDecl :: SC in CONEX grammar. It is assigned to none of the concept semantics in the form of:

<!ELEMENT StructDecl EMPTY>

Simply speaking, it is a meaningless symbol [2], which relates to nothing of concept in sign model [3]. The only structural meaning, which it may refer to, is that it is ready to be struc-turally modelled for evolving to a more complex meta-structure. Based on this simplest struc-ture, further syntactic rules are developed to describe XPM syntax applicable to CONEX Pro-totype implementation.

(2) Connotative Meta-Structure: Given a set of tokenised meta-structures StructDecls, a connotative meta-structure is an implementation of classifier structure for “classifierStruc-ture(StructDecl(…(…StructDecl)))”.

    <!ELEMENT Root (StructDecl*)>
    <!ELEMENT StructDecl (StructDecl*)>


where each StructDecl is a concept structure declaration. The Root introduces a catalogue root. The recursively connoted StructDecl describes a concept structure hierarchy.

Connotative meta-structure is open to spawn new meta-structures and conforms to the evolvability requirement of the CONEX approach. It satisfies the indexation of IID (Definition 4.6) and makes StructDecls indexable in implementation according to a document sequence. This syntax naturally constructs a catalogue structure, which could be structurally modelled to any catalogue frameworks. For example, UNSPSC [4] product standard can be modelled as a four level meta-structure:

 <root>                 <!-- catalogue -->
   <structDecl>        <!-- segment -->
     <structDecl>      <!-- family -->
       <structDecl>    <!-- class -->
         <structDecl/> <!-- commodity -->
       </structDecl>
     </structDecl>
   </structDecl>
 </root>.


(3) Denotative Meta-Structure: Given a tokenised meta-structure StructDecl, a denotative meta-structure is an implementation of concept structure c(s1, …, sn) of CONEX grammar such that:

   <!ATTLIST StructDecl
             (ElemStructDecl PrimStructDecl #REQUIRED | #IMPLIED)n> , [1]


where if the occurrence of ElemStructDecl is zero, then the StructDecl has no denotation and is degenerated as a tokenised meta-structure. The number of ElementStructDecls deter-mines the complexity of interior concept structure (see Section 5.2.1.1.1). For example, a general denotative concept structure is:

<!ATTLIST concept
          iid ID #REQUIRED
          annotation CDATA #REQUIRED
          link CDATA #REQUIRED
          option CDATA #IMPLIED>


where the iid is a unique concept identifier in a denotative meta-concept structure.

(4) Meta-Structure Model: Given a meta-structure in the form:

<!ELEMENT StructModelDecl EMPTY>
      <!ATTLIST StructModelDecl
(ElemStructDecl PrimStructDecl #REQUIRED | #IMPLIED)n
     ModelReference CDATA #IMPLIED>,


then it is said to be a meta-structure model if and only if the following structure (called be-modelled structure) is derived from the above meta-structure:

<!ELEMENT StructBeModelledDecl EMPTY>
      <!ATTLIST StructBeModelledDecl
(ElemStructBeMod PrimStructBeMod #REQUIRED | #IMPLIED)n>
     ModelReference CDATA #IMPLIED>.


The meta-structure model is a special type of implementation of the substructure and super-structure of s :: s’  of CONEX grammar in terms of be-modelled and model. The key to the model and be-modelled relation is that s and s’ should have same signature and s has a model reference to s’. This implementation is very important in establishing meta-structure equivalence rela-tionship between heterogeneous component EPCs. It allows two syntactically equivalent XML structures to be used interchangeably regardless of the heterogeneity in the string names of element nodes and attribute nodes. More specifically, two meta-structures are struc-turally isomorphic if they share the same structure signature (that is, one element with the same number of typed attributes) and the be-modelled structure has a reference to its model. For instance, the following two concept structures have the model and be-modelled relationship, which are structurally isomorphic.

<!Element concept EMPTY>
<!ATTLIST concept
          iid ID #REQUIRED
          annotation CDATA #REQUIRED
          link CDATA #IMPLIED>
<!ELEMENT 概念 EMPTY>
<!ATTLIST 概念
          标识 ID #REQUIRED
          注解 CDATA #REQUIRED
          地址 CDATA #IMPLIED
          引用 CDATA #IMPLIED>
<!--引用 references to the value of “link” of “concept”-->


The above meta-structure model permits an XML processing system to be modelled on an-other heterogeneous system, which increases the document reusability. Any two adjacent component EPCs can build model and be-modelled relationships for allowing model reuse if and if both implement the LINK elementary structure in a model. The value of LINK is the reference point, which states the be-modelled structure obeys the referenced structure. For example, when an operation of “概念” receives the information of concept(LF111, fridge, www.boo.com : structure : concept) from “concept”, “概念” will know that “concept” is the structure name that “概念” has referenced, “LF111” is the value of “标识”, and “fridge” is the value of “注解”.

(5) Meta-Concept: Given a meta-concept structure, for example:

<!Element concept EMPTY>
<!ATTLIST concept
          iid ID #REQUIRED
          annotation CDATA #REQUIRED
          type CDATA #REQUIRED>,

then the structure is said to be conceptualised if and only if all attributes of ATTLIST (i.e. a n-tuple of elementary structures) are instantiated by instance data (i.e. instances of primitive structures). A meta-structure conceptualisation is a meta-concept declaration such that for all elementary structures in a concept structure, t : s in CONEX grammar. For example, the following is a concept:

<concept iid = “1.52.14.15.1” annotation = “domestic re-frigerators” type = “product”/>.

In general, a meta-concept has the form of  “<concept x1 = “x1Val” … xn = “xnVal”/>”. A conceptualisation process is a collaboration process for creating a meaningful concept to in-stantiate a meta-structure.

(6) Reified-Concept. A reified-concept is an instance of the following DTD:

<!ELEMENT StructDecl (#PCDATA)>
  <!ATTLIST StructDecl (ElemStructDecl PrimStructDecl #REQUIRED | #IMPLIED)n>.


Specifically, a reified-concept has the form of “<concept x1 = “x1Val” … xn = “xnVal”> #PCDATA </concept>”. Its corresponding CONEX grammar is “X <= concept(x1Val, …, xnVal) : concept(x1, …, xn)”. The difference between a meta-concept and a reified-concept is that a reified-concept includes a text node, while a meta-concept has no text node.

The above constructs constitute the core syntax of XPM representation specification.

Reference

  1. Bray, T., Paoli, J. and C. Sperberg-McQueen, Extensible Markup Language (XML) 1.0, W3C Recommendation 10-February-1998, http://www.w3.org/TR/1998/ REC-xml-19980210.
  2. Chandler, D., Semiotics for Beginners, http://www.aber.ac.uk/media /documents/ S4B/semiotic.html
  3. Saussure, F., Course in General Linguistics, McGraw-Hill Book Company, 1966.
  4. UNSPSC, www.unspsc.org.
------------------------------------------------------------------------
[1] For simplicity, the ( )n is used to notate “zero-to-N” differently named attributes, which is applicable to other DTD notations in this thesis. In actual DTD modeling, it should strictly follow the rules of XML DTD to enumerate the named attributes. The “#REQUIRED | #IMPLIED” should also be separated, following XML 1.0 recommendation.

[2] Reference for citation: "Jingzhi Guo, Syntax of XML Product Map, Section 6.2.1.1 of Integrating Ad Hoc Electronic Product Catalogues through Collaborative Maintenance of Semantic Consistency, PhD Thesis, Griffith University, Australia, 2004".

Home Teaching Research Publication Demo Resource