XML/EDI Repositories

1
XML/EDI Repositories

• Martin Bryan
• The SGML Centre
• www.sgml.u-net.com

2
What is XML/EDI?

• The use of XML for business-to-business
  electronic data interchange
• Based on traditional EDI processes and existing
  semantic definitions
• UN EDIFACT, ANSI X12 or domain specific
• May or may not be based on existing message
  structures
• May be based on existing message implementation
  guidelines

3
On XML DTDs and Schemas

• A reminder of Heikki Mannila's point
• "One of the problems of using XML as a data
  exchange format is that DTDs only tell you the
  name of the element or attribute (and possibly
  the names of some of the permitted values of
  attributes). They say nothing about the meaning
  of this name."
• Micorsoft Corp.

4
The possible roles of a repository

• To identify existing message definitions that
  meet user requirements
• Need to know the purpose of the message
• To allow users to design new messages by making
  minimal modifications to existing messages
• Customising or extending existing message
• To allow users to design new messages using
  components that are associated with known
  processes

5
Underlying model
2. Transaction
Organization 1
Organization 2

1. Establish relationship between business process
   and repository definitions

1. Identify relationship and any associated processes

Repository
6
Operations

• a) Identify required business process
• b) Identify equivalent process in repository
• c) Record repository keys in DTD
• a) Transmit DTD
• b) Transmit messages referencing the DTD
• a) Use repository keys to retrieve information
  about relevant process
• b) Identify equivalent local business process

7
Keeping the relationships simple

• Only pass the reference to the repository
  information, not the information itself
• Could be in separate file, using links
• lt!ELEMENT My-thing (PCDATA)gt
• lt!ATTLIST My-thing
• href CDATA FIXED
• 'handle//www.standards-body-x.org/
• repository-y/glossary-zid(A003459)'
• Simple-XLink-Attributesgt
• Does not work for link elements

8
The overall process
As defined in the XML/EDI Group's Repository
white paper
9
Who's repository is it anyway?

• Handle server does not need to store the DTDs
• Acts as standard reference method for multiple
  repositories
• Repositories can serve multiple communities
• Common API allows repositories to report
  information in standardized format

10
What language must I use?

• Repositories must become multilingual
• Multilingual glossary definitions
• Single identifier for all definitions
• Object naming up to user communities
• Objects can conform to multiple definitions
• Multiple identifiers act as bridge between
  repositories
• Requires use of extended links

11
Multiple ownership

• Need to allow for ownership hierarchies
• UN EDIFACT
• Industry-based European Expert Groups
• Regional groupings
• National applications
• Company agreements
• Naming may differ between owners
• Element names need not reflect semantic names

12
How can we handle this?

• Use XML namespaces
• lt!ATTLIST my-local-name
• xmlnsEDIFACT CDATA "http//www.unece.org/D99A"
• xmlnsTransport CDATA "http//www.EEG2.eu/IFT"
• EDIFACTSegment CDATA FIXED "FTX"
• TransportDataSet CDATA FIXED "Notes"
• Use XML links?
• Simple link add-ons or extended links?
• Referencing externally managed pointers to a set
  of relationships between semantic sets?

13
Benefits of using a repository

• Self documenting XML DTDs
• Code reusability
• Sharing presentation rules
• XSL modules can be based on the repository link
  attribute rather than the local name of the
  element
• Message elements with the same identifiers are
  candidates to share business processes
• Can automate process of associating message
  subsets with processes

14
Identifying existing DTDs

• Need to interrogate metadata of DTD
• Who generated it, when and why?
• What types of processes is it defined for?
• Can it be modified or subsetted?
• Can its code lists be redefined?
• What documentation is associated with it?
• Need to be able to reference the source
• Using public and system URLs
• How will updates be managed?

15
Existing DTD repositories

• XML.org Specifications Catalog
• Organization for the Advancement of Structured
  Information Standards (OASIS)
• http//www.xml.org/xmlorg_catalog.htm
• DTDs and Schemas
• XML.org Recommendations
• CommerceOne Common Business Language
• DataChannel Portal Markup Language
• Microsoft's BizTalk

16
BizTalk

• BizTalk Framework mostly concerned with routing
  of messages
• BizTalk defined in terms of Microsoft's
  proprietary XML-Data schema format
• Little control over who submits schemas
• Categorization currently limited
• Construction, Education, Finance and insurance,
  Information, Manufacturing, Professional/Scientifi
  c and Wholesale

17
BizTalk schema metadata
18
BizTalk View Schema (XML-Data)
19
BizTalk documentation
20
BizTalk sample
21
Modifying existing DTDs

• Have parameter entities been used to allow model
  customisation?
• Removal of optional components
• Defining new mandatory components
• Have parameter lists been used for enumerated
  code lists?
• Can you reference an external code list?
• Can you subset internal code lists?

22
Creating new DTDs

• How can I determine which repository objects to
  use?
• Classification of object purpose within
  repository glossary
• Metadata or internal documentation?
• Identifying where previously used
• What can I call it?
• What is the referenceable unit in the repository?
• Need to be able to reference all subelements and
  their attributes

23
Common Business Language

• CBL 2.0 Reference Guide
• http//www.marketsite.net/xml/cbl/reference.html
• Limited set of messages
• Purchase Order/Purchase Order Response
• Order Status Request/Order Status Result
• Invoice
• Availability Check Request/Result
• Price Check Request/Result
• Cataloguing and Pricing information exchange

24
CBL Repository
25
CBL SOX schema
26
What I really really want

• XML DTDs and/or XML Schemas
• Elements for defining processes
• Personal Identification
• Family name, given name, initials, title
• Exchange addresses
• Name/number, street, town, region, country
• Identify product requirements
• Item id, item name, quantity, delivery date
• Links to existing processes
• APIs for local databases
• APIs for process controls

27
and exchangeable semantics

• ISO 11179 Data element specifications?
• Due for updating
• Standardized ontology descriptors?
• Synonyms (but not antonyms and homonyms)
• Hypernyms and hyponyms (broader and narrower
  terms)
• Holonyms and meronyms (wholes and parts)
• XML/EDI group's repository API?

28
Tell me more

• XML/EDI Group Repository Initiative
• http//www.xmledi.org
• Read their Repositories White Paper
• http//www.xmledi.com/repository/xml-repWP.htm
• Join their Repositories working group
• http//www.xmledi.com/repository
• CEN/ISSS Defining and Managing Semantics and
  Datatypes project group
• http//www.cenorm.be/isss/Workshop/ec/projects.htm