RE: [ubl-ndrsc] Tag structure discussion

["Schuldt, Ron L" <ron.l.schuldt@lmco.com>]

Title: RE: [ubl-ndrsc] Tag structure discussion

Mark,

 

The value of a structured ID is that it allows one to compare/align the IDs of data from different systems (particularly legacy systems) using the power of a computer to find perfect matches in some cases or close matches in other cases.

 

For example, if one company mapped their legacy data for <SupplierEnterpriseIdentifier UID="y.3_8"> and another company mapped their legacy data for <SupplierEnterprisePuchaserAssignedIdentifier UID="y.3_2.35.8"> a parser could flag the possibility that the data were in fact the same - based on the fact that y.3_8 is contained within y.3_2.35.8 and by applying the rules of UDEF where both contain the same representation term (identifier) and both contain the same object term (supplier enterprise) - there is a possibility that the two are semantically equal. Without intelligence in the UID, an automated comparison/alignment would be much more difficult. A major task when two new trading partners are attempting to conduct business with each other is the effort associated with aligning data between their differing systems.

 

I would argue that a important part of the audience that will make ebXML fly are the large (Fortune 1000) companies and government organizations that have substantial investments in legacy systems. The vast majority of these large companies and government organizations will be relying on their legacy systems and will need to find ways to simplify the comparison/alignment of those systems to those of their trading partners. I suggest that structured IDs such as offered by UDEF is the way to simplify the alignment or comparison process.

 

Ron Schuldt

 

 

 

-----Original Message-----
From: CRAWFORD, Mark [mailto:MCRAWFORD@lmi.org]
Sent: Wednesday, December 19, 2001 5:00 AM
To: ubl-ndrsc@lists.oasis-open.org
Subject: RE: [ubl-ndrsc] Tag structure discussion

>After all, why are ebXML
> elements given
> unstructured UIDs?

In the ebXML architecture, the work of matching the UID to the tag/core component/bie occurs in the registry so structure is unimportant.  You will either know the value of the item you are looking for, or not.  If you have structured tags, then it is much more likely that you will get a match.  I don't see how having a structured ID number will in any way ease the lookup or cross reference.  Just as you must follow the rules for structured tag (Object Class, Property Term, Representation Term), so you must follow the rules for developing a structured UID.  In both cases, the ability to follow the rules is important to arrive at something that can be searched for.  Neither has an advantage over the other, other than structured tags carry a semantic value that structured UID's do not and so are more intuitive and less dependent on a mathematical conversion routine. And remember, a large part of our audience will be folks that must rely on what they see to convey semantic clarity.

I wholeheartedly agree that structure of the tag name is important, especially when we consider that one of the principle aims of UBL is to develop a standard business XML vocabulary (See Jon's recent presentation and the prominence this theme plays).  Unstructured, or semi-structured tags just do not accomplish this goal. However, the argument for structured ID's just hasn't convinced me they are necessary.  It seems to me it is a lot of complication that adds an unnecessary layer on top of  the structured tag name approach.

Mark