A DOCUMENT RULES DESCRIPTION LANGUAGE BASED ON

FEATURE LOGIC FOR XML DOCUMENT EXCHANGE

Makoto Imamura

, Yasuhiro Takayama

, Yasuhiro Okada

and Norihisa Komoda

Mitsubishi Electric Corporation,

Osaka University

Keywords: Document Exchange, XML Validation, XML Transformation, Schema Description, Logic Programming.

Abstract: We propose a minimum Document Rules Description Language DRDL for XML validation and translation

in web-based document exchange. DRDL has small syntax and simple semantics so that it can ease to reuse

document rules. Furthermore, DRDL can describe mapping constraints from XML tree structure to table

structure for web input-form generation. The technical feature of DRDL is to extend and reinterpret a

Feature Logic, which has been used for representing linguistic knowledge, in order to allow existential and

universal quantifiers over list-value to cope with XML. DRDL has already been applied to real systems

such as elevator design support, Web-EDI, government-to-business and facility management.

1 INTRODUCTION

As XML formats have been widely adopted for a

standard, interoperable document exchange format,

XML validation and translation become common

and critical component for document exchange both

within and across enterprises. In developing web

based XML document exchange systems, it is a

problem how to efficiently make data validation

function and table style form layout.

XML standard technologies, such as XML

schema, XSLT and XQuery, have been developed to

increase expressive power for document validation,

transformation and query. However these standards

are too huge to easily master and are too heavy to be

used in some applications.

In this paper, we propose a simple language

DRDL which has the following features to describe

semantic constraints in document rules.

1. DRDL has small syntax and simple semantics so

that it can ease to reuse document rules, such as

input-form check for elevator specification or

Web-EDI, by translating rules to other languages

such as JavaScript.

2. DRDL can describe table mapping constraints

from XML tree structure to table structure for

web input-form.

The remainder of this paper is organized as

follows: Section 2 presents the requirements for

DRDL. Section 3 defines the syntax and semantics

of DRDL. Section 4 presents an DRDL-based

framework, which consists of an editor, rule translators

and a processor. Section 5 discusses related works,

while section 6 presents conclusions and open

issues.

2 REQUIREMENTS

We present here the requirements of expressive

power and the development productivity for DRDL.

2.1 Expressive Power

In the industrial and business document exchange

systems, document validation and transformation are

indispensable functions for smoothly processing the

exchanged document data to prevent errors. DRDL

needs to have sufficient expressive power to

describe the following functions to treat cross-field,

cross-field-structure cross-field-type and table

mapping constraints, which the first version of XML

Schema can not handle (Daum, 2003).

a) Cross-field constraint processing

To validate constraints across the contents of

elements or attributes in XML documents. For

example, to check that each element <A> is equal to

the sum of each element and <C> in their

respective orders for multiple <A> elements.

Imamura M., Takayama Y., Okada Y. and Komoda N. (2007).

A DOCUMENT RULES DESCRIPTION LANGUAGE BASED ON FEATURE LOGIC FOR XML DOCUMENT EXCHANGE.

In Proceedings of the Second International Conference on e-Business, pages 71-77

DOI: 10.5220/0002111800710077

 SciTePress

To assign the value to an element which has been

computed for the contents of elements (e.g. the sum

of multiple elements).

b) Cross-field-structure constraint processing

To validate constraints between the structure of an

element and the content of an element in XML

documents. For example, to validate whether the

multiplicity of element <A> is equal to the content

value of element .

To change the structure of an element to satisfy a

cross-field-structure constraint. For example, to add up

the <A> elements such that the multiplicity of element

<A> is equal to the content value of element .

c) Cross-field-type constraint processing

To validate constraints between the type of an

element and the content of an element.

To assign a value to element <A> depending on

the type of the content of element .

To validate if data in an XML document are

consistent with those in RDB (Relational-Database).

d) Table Mapping Constraint Processing

To describe mapping constraints from XML tree

structure to table structure for input form generation.

2.2 Development Productivity

In order to apply DRDL to developing industrial

systems, the XML application development

frameworks need to have the following

requirements.

a) Development efficiency

Firstly, from the development efficiency point of

view, the frameworks have the following

requirements for easily revising document

processing functions with schema upgrade:

1. Editability of document rules

Document rules need to be edited not only by the

programmers in an information system division, but

also by the end-users in a related business division.

2. Reusability of document rules

Document rules need to be easily shared among

organizations participating in the document

exchange, and to be utilized easily by each

organization’s business sub-systems.

3. Replacability of document rules

Document rules need to be easily replaced in the

distributed sub-systems which use a common

document format.

b) Connectivity to other functions

Document rules need to be processed with other

functions in the application system. For example,

rule validation functions should be able to be

communicated with a document view controller in a

client-side document input tool.

3 XML DOCUMENT RULES

DESCRIPTION LANGUAGE

We define here DRDL which has the expressive power

stated in 2.1. 3.1 presents design principles. 3.2 and 3.3

defines the syntax and the semantics of the core part of

DRDL which is called XML constraint language

(XCL). 3.4 presents DRDL as the extension of XCL

with optional functions for application systems.

3.1 Design Principles

The main design principle of DRDL is to define a

minimum language needed to describe semantic

constraints in Web-form based XML document

exchange. To define DRDL, we use the concept of

unification grammar formalism (Shieber, 1986) which

describes natural language syntactic and semantic

constraints. Description in unification grammar

consists of production rules described by context-free

grammar, and feature constraints described by feature

logics whose typical example is (Smolka, 1992).

Production rules correspond to document structure

definition described by DTD or XML schema, and

feature constraints correspond to semantic constraints

described by XSLT or XQuery. We extend the syntax

of a feature logic and reinterpret the semantics of it in

order to cope with XML.

Table 1: Analogy with feature logic and XML processing.

XML processing Feature logic

Validation Transformation

path in feature

structure /

location path in XML structure

and

∧

and sequential statement

implication

⇒

implication if statement

universal

quantification

∀

for-each statement

existential

quantification

∃

cardinality check

of elements

addition or deletion

of elements

equation = equality check

unification

(variable assignment)

sort : type check (no correspondence)

subsumption ⊆

upper compatibility

of schema

(no correspondence)

ICE-B 2007 - International Conference on e-Business

The main syntactical extension is to introduce

quantifiers over list-value. Semantic reinterpretation is

to introduce validation interpretation for XML

document validation functions in 2.1, and assignment

interpretation for XML document transformation

realized by the assignment functions in 2.1.

A core idea of DRDL which will be stated in the

following subsections is to use semantical analogy

between a feature logic and XML processing in table 1.

3.2 Syntax of XCL

We define in this subsection the syntax of XCL, the

core part of DRDL, and a term and a formula in

XCL like those in a first-order language.

Definition: XCL term

An XCL term, which denotes a list of XML nodes, is

defined as follows:

(1) A constant (a list of XML nodes) is an XCL term.

(2) A location path in XML is an XCL term.

Definition: XCL formula

An XCL formula is inductively defined as follows:

(1) XCL comparison formulae

If s iand t are XCL terms, then s = t, s

≠

t, s

≤

t, s

≥

s>t and s<t are XCL formulae.

(2) XCL type-constraint formula

If s is an XCL term and t is a data type, then s:t is an XCL

formula.

(3) XCL logical formulae

If F and G are XCL formula, then F

∧

G, F

∨

and F

⇒

G are XCL formulae.

(4) universally quantified XCL formula

If x is a variable, p is a location path, and F is a XCL

formula, then

∀ x

∈

p. F is an XCL formula.

(5) existentially quantified XCL formula

If x is a variable, p is a location path, n is a non-negative

integer or an expression whose evaluated value is a non-

negative integer, and F is XCL formula, then

∃

∈

p s.t.

(count(.)=n).F is an XCL formula. In (count(.)=n), the

symbol “=” may be

≠

≤

≥

or > and the symbol “n” may

be an arithmetic expression.

3.3 Semantics of XCL

We present the operational semantics of each

formula stated in section 3.2.

1. XCL Comparison Formulae

“s = t” is evaluated as follows in validation

interpretation. If the evaluation value of “s” is equal

to the evaluation value of “t” in the sense of list,

then it is “true”. Otherwise it is “false”.

“s=t” is evaluated as follows in assignment

interpretation. An evaluation value of “t” is assigned to

the position that is addressed by a localtion path “s”. This

evaluation procedure corresponds to a DOM-API “set

node value” which assigns a value to a node in a DOM

tree.

2. XCL type-constraint formula

“s: t” is evaluated as follows in validation

interpretation. If the evaluation value of s belongs to

data type t, then it is true. Otherwise, it is false.

The evaluation of “s: t” is not defined in

assignment interpretation.

3. XCL Logical Formula

XCL logical formulae in validation interpretation are

evaluated as those in propositional logic. XCL logical

formulae in assignment interpretation are evaluated as

follows: For F

∧

G , F is firstly evaluated and

secondly G is evaluated; For F

∨

G , only F is

evaluated; For

F, it is reduced to the normal form

F’, in which all

s are attached to comparison

formulae by rewriting with tautology in first-order

predicate logic, and then F’ is evaluated; For F

⇒

if the evaluated value of F in the validation

interpretation is true, G is evaluated. This evaluation

procedure corresponds to the “if” statement in XSLT.

4. Universally quantified XCL Fformula

In respect of validation interpretation,

∀

∈

p. F is

evaluated as follows. Firstly, x in formula F is

substituted by each element of a list value, which is

obtained by evaluating a location path p, and we can

then obtain formulae Gs which are the variants of F.

The number of formulae Gs is equal to the

cardinality of list values of a location path p. If all

the formulae Gs are evaluated and all the values of

those are true, the formula F is true; otherwise, it is

false.

In respect of assignment interpretation,

∀

∈

p. F

is evaluated as follows. Formulae Gs as the variants of

F are obtained by the substitution of x in F in the same

way as that used in validation interpretation. Each

formula G is evaluated in the same order as that in the

list value of p. The order of an element in the list value

of p is naturally defined by the order of a value in an

XML document. This evaluation procedure

corresponds to the “for-each” statement in XSLT.

5. Existentially quantified XCL formula

In respect of validation interpretation,

∃

∈

s.t.(count(.)=n). F is evaluated as follows. Formulae Gs

as variants of F are obtained by the substitution of x in

F in the same way as

∀

∈

p. F. If the number of

these formulae is equal to the evaluation value of “n”

A DOCUMENT RULES DESCRIPTION LANGUAGE BASED ON FEATURE LOGIC FOR XML DOCUMENT

EXCHANGE

and at least one of the evaluation values of G is true,

then it is true. Otherwise it is false. This evaluation

procedure corresponds to minOccurs and maxOccurs in

XML Schema, which check the multiplicity of

elements. Quantified XCL formulae are much more

powerful than those in XML schema, because “n” can

be an arithmetic expression.

In respect of assignment interpretation,

∃

∈

p s.t.(count(.)=n).F is evaluated as follows.

XML elements are added or deleted until the number of

nodes, which are addressed by a location path p, is

equal to the evaluation value of “n”. Formulae Gs as

variants of F are obtained by the substitution of x in F

in the same way as that for

∀

∈

p. F. Only the first

formula in Gs is evaluated. This evaluation procedure

corresponds to an instruction for node operation in

DOM such as “appendChild” or “removeChild”.

3.4 Examples

We present in this subsection examples of the cross-

field, cross-field-structure and cross-field-type

constraints described in 2.1.

1. Cross-field Constraint

The formula shown in figure 1 describes a constraint

whereby, for each <p-list>, the <total> is equal to

the <sum> of each <price> in <product>s. An

example XML document is shown in fig. 2.

Validation interpretation achieves a validation

function for XML documents, while assignment

interpretation achieves a computation function

among cells in a table like a spread-sheet software.

∀

∈

/root/p-list. x/total = sum(x/product/price)

Figure 1: Cross-field Constraint with DRDL.

<p-list date＝"2002-11-01">.... </p-list> </root>

Figure 2: An Input XML Document.

2. Cross-field-structure Constraint

The formula shown in fig. 3 describes a constraint in

which a multiplicity of <device>s is equal to the

content of <device-number>. Fig. 4 shows an

example of an XML document as the presumed

input. Validation interpretation achieves a

cardinality checking function for the XML elements.

Assignment interpretation achieves addition and

deletion operations on XML elements which

correspond to DOM APIs such as “appendChild”

and “removeChild”.

An assignment interpretation of the formula in figure

3 for the input XML document in figure 4 is as follows:

Three <device> elements are added, consequently the

multiplicity of element <device> is equal to the content

of an element <device-number> which is 5.

∃

∈

/device-list/device s.t.(count(.) = /device-number).

Figure 3: Cross-field Constraint with DRDL.

<device-number> 5 </device-number>

<device-list><device> pencil </device>

<device> pencil sharpener </device></device-list>

Figure 4: An Input XML Document.

3. Cross-field-type constraint

An XCL type–constraint formula can define the

data-type of an element depending on the content of

another element, which cannot be described by an

XML Schema. The formula in fig. 5 describes a

constraint whereby, for each <product>, if the

<category> in the <product> is “communication-

facility”, the data type of the <number> of the

product is “communication-facility code”.

∀

∈

//product. ( x/category = “communication-facility”

⇒

(x/number: “communication-facility-code”)

Figure 5: Cross-field-type Constraint with DRDL.

3.5 DRDL

DRDL is an extended XCL which has the following

optional descriptions.

(1) Table mapping constraint ( stated in 2.1d) )

(2) RDB mapping constraint

(3) Vector expression constraint

In this subsection, we describe a table mapping

constraint which is the main feature of DRDL. Table

mapping constraint consists of a matrix constraint and a

connection constraint. Matrix constraint describes the

mapping between a location path in XML documents

and the row and column of a table. A connection

constraint describes the condition for tables to be

connected with those of rows or columns.

For example, the constraints in fig. 6 describe the

table mapping denoted by fig. 7. (1) and (2) in fig.6

describe mapping constraints from the left part of

doc1 to the left part of the table (tbl1). In a similar

way, (3) and (4) in fig.6 describes mapping

ICE-B 2007 - International Conference on e-Business

constraints for the right part of doc1 and tbl2. (5) and

(6) in fig.6 describe constraints to connect the left and

the right part of the table. Connection constraints can

describe dependency between the number of the

column or row of a table and that of the other table.

(a) Matrix constraint

tbl1 ! row = doc1 ! a1/b (1)

tbl1 ! column = doc1 ! a1/b/c (2)

tbl2 ! row = doc1 ! a2/bb (3)

tbl2 ! column = doc1 ! [ a2/bb/e, a2/bb/f ] (4)

(b) Connection constraint

count(a1/b) = count (a2/bb) (5)

horizontal_connect ([tbl1, tbl2]) (6)

Figure 6: Table mapping Constraints with DRDL.

<a2>

<bb>

</bb>

<bb>

</bb>

</a2>

12 11

13 14

<a1>

</a1>

1 3

4 5 6

Connection constraint

(

)(

)

[ Table]

Matrix constraint(1)(2)

[ Table (tbl1)]

[ XML document ( doc1 ) ]

[ Table (tbl2) ]

Matrix constraint(3)(4)

Figure 7: Table Mapping Example.

4 XML APPLICATION

DEVELOPMENT FRAMEWORK

a) Architecture of the DRDL Framework

DRDL Framework consists of a rule file, an editor, a

processor and a rule translator. Figure 8 shows the

whole structure of DRDL Framework.

DRDL Editor

DRDL File

Sy st em Dev el oper

RDB

DRDL Rule

Translator

DRDL

Processor

End -User

XML

Web

Browser

Client

Application

Input

Check Result

JavaScri

DRDL

Processor

Figure 8: Architecture of DRDL Framework.

1. DRDL Editor

In order to satisfy the requirement for “editability of

document rules (2.2a)1)”, we have developed a DRDL

editor for end-users who do not have specific IT

knowledge to enable them to make DRDL formulae.

The DRDL editor has a table-like GUI interface.

The left part of this GUI is for editing the document

schema, and the right part is for editing the document

content constraint or the XML-RDB mapping constraint.

2. DRDL Rule Translator

To fulfill the requirement for “reusability of document

rules (2.2a)2)”, we have developed DRDL rule

translators which transform DRDL formulae to

another programs or scripts for other XML

processors. For example, a translator can generate

JavaScript for web input forms from DRDL formulae.

These JavaScript are used for input checking in our

XSLT based XML input form (Imamura et.al., 2005).

3. DRDL File

To fulfill the required “replacability of document

rules (2.2 a) 3)”, we have introduced DRDL files to

describe document rules independently from

application programs. Sending and replacing DRDL

allows client XML tools to replace the document

processing function and server systems to replace

the XML transformation function.

4. DRDL Rule Processor

To fulfill the requirement for “connectivity to other

functions(2.2b))”, we have developed a DRDL

processor which can cooperatively process XML

documents sharing a DOM tree with other processes.

b) Application Systems of DRDL

The first version of DRDL framework was built in 1999

and has been applied to the following systems

(Imamura et.al., 2000). The number between parentheses

denotes the start year of the system operation.

(1) Elevator design support system (1999)

(2) G2B document-exchange system (2000)

(3) Web EDI system (2001)

(4) Facility management system (2002)

DRDL Edi t or

DRDL F i l es

Design Sheet

(

XML

)

DRDL Files

Des i gn

Support

Sy st em

Order Design Division

Sales Design

Support Syst em

Int ernet

Model Design

Division

Or d er

Inf o.

Des ign

Inf o.

Desi gn Sheet

Input Tool

DRDL

Pr ocessor

Specification Input

Tool

DRDL

Processor

Sales Design Division

Sales

Support

Sy st em

Sales Design Division

Figure 9: Elevator Design Support System.

A DOCUMENT RULES DESCRIPTION LANGUAGE BASED ON FEATURE LOGIC FOR XML DOCUMENT

EXCHANGE

For example, we show the elevator design

support system in Figure 9. Elevator design sheets

are sent from the sales design divisions to the order

design divisions. DRDL processors are used to

check whether the elevator design sheets satisfy the

document rules stipulated in the input conditions for

the design support systems. The distinct number of

XML elements for each design sheets is about 300,

and the total number of check rules is about 900.

DRDL is used effectively to describe table mapping

constraints depending on the number of floors and

lifts in the building. The DRDL framework

decreased 50 % of the time for creating Web-input

forms compared with an existing form builder.

5 RELATED WORKS

Various schema languages have been proposed in

order to describe document constraints stated in 2.1

without hard-coding (Murata et al., 2005). Famous

ones are Schematron (Jelliffe, 2004) and DSD

(Klarlund et al., 2002) as a pattern-based language.

DRDL is different form them in logic-based. DRDL

is expected easy to translate semantic constraint to

other languages by referring logic-based existing

formal specification works.

XSLT is a standard and popular language for

XML transformation. However its specification is

too huge to easily master, so there are some tools to

support XSLT generation (Tang et al., 2001). Table

mapping constraint description in DRDL is an

approach to specify the mapping pattern from XML

tree structure to table structure. In fact, DRDL has a

translation function from table mapping constraints

in DRDL to XSLT descriptions.

Recently, in the database community, declarative

and unifying approaches with document schema and

database query language have been proposed for data

integration and exchange. The BEA AquaLogic Data

Service Platform adopts data integration framework

based on XML Schema and XQuery to modelling and

accessing the variety of data source types such as

relational, Web service, function-based and file-based

(Reveliotis et al., 2006). In this paper, we discussed the

data integration and exchange from a document

processing and knowledge representation point of view.

6 CONCLUSIONS

We have proposed a minimum semantic constraint

description language DRDL which satisfies the XML

document-processing requirements of expressive

power, development efficiency and connectivity to

other functions needed in real application systems.

Interpretation of the DRDL formula realizes a

validation function for XML documents. Assignment

interpretation of DRDL formulae realize operations

on DOM trees, such as addition, deletion and value-

assignment like DOM-API, and also control

structures, such as the “if “ statement in imperative

languages and the “for-each” statement in XSLT.

Furthermore, DRDL provide table mapping constraint

for Web-input form generation.

Open issues are the following.

(1) Logic programming with terms as XML elements

We have treated the assignment interpretation in a way

analogous to popular imperative languages such as

Java or C for usability and performance. From the

theoretical point of view, however, it is important that a

determination procedure for the satisfiability of XCL

formulae should give a unification algorithm between

XCL formulae (Smoka, 1992). Replacing a term in Prolog

with an XCL formula allows us to obtain a new

constraint logic programming (Jaffar et al., 1994)(Mukai,

1991). This logic programming language is an

alternative XML transformation language to XSLT.

(2) Decision procedure for satisfiability of subsume

relation of XCL formulae.

(3) Comparison with other declarative XML constraint

description languages such as Xcerpt (Schaffert, 2004)

and Relational.OWL (Laborda et.al., 2005).

REFERENCES

Daum,B. 2003, Validation beyond XML Schema,

Modelling Business Objects with XML Schema,

Morgan Kaufmann Publishers, pp.323 -362.

Imamura, M. et al. 2000. Metadata representation for

Internet-based XML application from business to

government, in proc. of 7th International Conference on

Parallel and Distributed Systems Workshops, pp.387-392.

Imamura, M. et al. 2005. An XML Input Form Generation

Method Based on a Tree-table mapping Model,

Transactions of Information Processing Society of

Japan, Vol.46 No.12, pp3066-3077.

Jaffar,J. and Maher,M.J. 1994. Constraint Logic

Programming: A Survey, Journal of Logic

Programming 19, 20 pp.503-581.

Jelliffe, R., 2004. The Schematron, http://xml.ascc.net/

schematron/

Klarlund, N., Anders Møller,A., Schwartzbach ,M.I. 2002,

The DSD Schema Language, Automated Software

Engineering 9 3, 285-319

ICE-B 2007 - International Conference on e-Business

Laborda, C. P. and Conrad, S. 2005. Relational.OWL, in

proc. of Second Asia-Pacific Conference on

Conceptual Modelling (APCCM2005), pp.89-96.

Mukai, K. 1991. Constraint Logic Programming and the

Unification of Information, Doctoral Dissertation, Dept.

of Computer Science, Faculty of Engineering, Tokyo.

Murata, M., Lee, D., Mani, M., Kawaguchi,K, 2005.

Taxonomy of XML schema languages using formal

language theory, ACM Transactions on Internet

Technology (TOIT) Volume 5 ,Issue 4, pp.660-704

Reveliotis, P., Carey, M., 2006. Your Enterprise on XQuery

and XML Schema: XML-based Data and Metadata

Integration, in proc. of 22nd International Conference

on Data Engineering Workshops, pp80 – 89.

Schaffert,S. 2004, Xcerpt: A Rule-based Query and

Transformation Language for the Web,

http://edoc.ub.uni-muenchen.de/archive/

Shieber, S. M. 1986. An Introduction to Unification-based

Approaches to Grammar, CSLI Lecture Notes

Number4, Stanford University

Smolka, G. 1992. Feature Constraint Logics for

Unification Grammars, Journal of Logic

Programming, New York, PP51-87.

Tang,X. and Tompa,F.W. 2001. Specifying Transformations

for Structured Documents, Proc. of the 4th International

Workshop on the Web and Databases, pp.67-72

A DOCUMENT RULES DESCRIPTION LANGUAGE BASED ON FEATURE LOGIC FOR XML DOCUMENT

EXCHANGE