It is interesting (and frightening to some) to look at how technology influences language. My latest observation in this regard is how “canonical” — as used in computer science — has progressed from adjective to noun in an increasing number of discussions around service oriented architecture (SOA). I notice that “canonicals” passes muster in my spell checker, but I’m thinking that may have to do with the word’s ecclesiastical origins.
Wikipedia explains the modern etymology of canonical by noting that the field of computer science borrowed the term from its use in mathematics to describe something in its simplest form without loss of generality or something reduced to a basic or standardized form (Wikipedia doesn’t yet recognize the word as a noun). The earliest computer-science usage of canonical I personally recall was in connection with so-called “canonical SGML,” a precursor to XML.
“Canonical” in the context of an increasing number of SOA discussions is used in referring to “common information models” or “CIMs.” I believe credit for coining the usage in this context (“canonical model,” “canonical data model,” etc.) has to go to the folks at the Open Applications Group (OAGi), which began its work in 1994 with the recognition that a common information model for enterprise integration was something that prior EDI initiatives had not yet delivered. So while the terms “canonicals” and “CIMs” appear with increasingly frequency within the discussions about “making SOA work,” it is worth noting that the canonical model concept has long been part of the parlance of the OAGi community.
In a prior post, I introduced the idea of HR-XML 3.0 as a leap forward towards a canonical model for HR. If there ever was a domain in need of a canonical model, it is this thing we call human resources. I’ve frequently told the story about a conversation I had years ago when HR-XML was in formation with Laura Walker, then director of the Organization for the Advancement of Structured Information Standards (OASIS). Laura’s logical, dispassionate reaction to my description of what HR-XML might be was that it sounded more like two or three organizations rather than one. That is the crazy thing about HR — very disparate processes and activities are bound together by the need for common (or compatible) views of things like “resource,” “employee,” “competency,” “time worked,” etc.
I realize that there are people both within and outside the HR-XML community who are still trying to get their heads around exactly what HR-XML 3.0 represents. If you had to put it in a nutshell, HR-XML 3.0 brings together HR-XML’s almost a dozen sub-domains (things like recruiting, hiring, payroll, health and welfare benefits, savings and stock plans, screening and background checks, assessments and performance management) under a common, integrated data model. Some might ask if that wasn’t what HR-XML already had. While HR-XML’s 2.* series of releases reused common content (what we called Cross Process Objects or CPOs), the various sub-domains weren’t truly integrated within one model. That is not to say we did anything wrong! Really, I can’t imagine what methodology we might have used to bring our volunteer sub-domain experts in very diverse fields together to build the common information model in a top-down fashion.
Bringing everything together within one model sounds big — It is! This is a big job, but one that is built on top of existing work (that of HR-XML, OAGi, and groups like UN/CEFACT). While the model will be “big,” if you delve into the details, I think you’ll find the new architecture is very scalable and flexible. Just about every component in the library is globally scoped so implementers will have a lot of choice and flexibility compared to HR-XML 2.* schemas that were a bit more hierarchical and monolithic in design. Implementers will be able to use HR-XML 3.0 business object documents (BODs) as is or extend them or “roll their own” using the rich library of global components.
So 3.0 is simply a logical progression of the earlier work. Now that we have specialized HR sub-domains richly described, we can move beyond mere sharing of common objects to something more integrated — something much more like a canonical model for HR. If you look at current discussions regarding the challenges many enterprises have in implementing service oriented architectures, a “canonical” for HR may be coming along just at the right time.
Learn More in Atlanta
Want to learn more about HR-XML 3.0? There will be no better place than at HR-XML’s March 3-4 meeting in Atlanta. Register and make your hotel reservations this week!