The Information Framework Structure – Top-down or Bottom-up?

I was home for the what seem to be long-ago holidays, but took a break from doing a blog since many of you were on an extended holiday from work.  So, my food adventure was home-based.   Jeannie (my wife) and I had a crab-fest rather than a traditional Christmas dinner.  Sara, the cat, even had a taste.  Simple steamed Alaskan king crab legs and corn on the cob only takes about ten minutes…the hardest part was shelling the crab after it was cooked.  One requirement is lots of drawn butter on the side for dipping.  Yummy!

A traditional assumption about the Information Framework’s (SID) structure is that it was developed from the top down.  When I am out teaching this assumption is often posed as a question about the origins of the structure.  Well, just like my Christmas crab-fest dinner, tradition was not followed in the development of the Information Framework structure.  The structure was developed using top-down, bottom-up, as well as middle-up-and-down approaches.

But, the approaches all have to do with the concept of affinity.  Much like popular affinity groups on the internet today, we like to define clusters of “things” that are closely related to each other based on one or more characteristics, while being loosely related to things in other clusters.  For example, a large affinity group may be made up of book lovers, while smaller groups are based on the primary subject of interest, such as science fiction or mystery.

Believe it or not there a technique called affinity analysis that can be used to quantitatively measure the “strength” of relationship between “things”, even though many of us intuitively perform it.  Here is a link for those of you who have a mathematical interest in the subject…

http://it.toolbox.com/blogs/enterprise-solutions/affinity-analysis-14726

A key point here is that affinity analysis can be conducted on just about any type of object.

From an information model perspective, there are variations of affinity applied to the entities that make up the model.  One measures affinity using associations (relationships) and the other measures affinity using processes’ actions on entities.

So you don’t think I’m always a top-down kind of guy, let’s start with the bottom-up approach.  This assumes that there is an information (or data) model that has been developed.  And, that there is a desire to check the correctness of any intuitive grouping and/or to group entities that haven’t been grouped at all.  One of the TM Forum members who originally contributed to the SID did this, which provided one view of a possible SID structure for Aggregate Business Entities (ABEs) and entities.  Domains came later.

Another member had used somewhat intuitive, process-focused affinity analysis, using the Business Process Framework (eTOM) Level 2 core processes.  This could be viewed as a middle-up-and-down technique.  The key objects acted on by Level 2 processes were assumed to be an ABE.  Development/grouping of the entities proceeded from this “mid-point” down.  Domains came later.

From a top-down approach, the eTOM already had a set of key concepts that were integral to it, meaning that there were large groups of closely related processes that support such concepts as Customer and Product.  And, these exist in almost every enterprise-wide process model, not just the eTOM.  This became the eventual “top” of the SID framework.  It’s interesting to note that the original “bottom-up” set of domains did vary somewhat from the eTOM concepts.

The take-away from this is that, no matter what approach is used, affinity analysis can be used at any stop along the way to develop or confirm the structure of any information framework.

I’ll be on the road again next week for Team Action Week Paris 2011.  Hope to see many of you there.