你使用什么模型处理XBRL？ - Charlie

Something occurred to me as I was working on something. XBRL Dimensions has
always gotten a lot of heat. "Too hard to understand". "My information is not
dimensional." There are other such statements, you have probably heard them.

But let's turn the question around and look at it from the opposite perspective.
If you don't like XBRL Dimensions or the multidimentional model which XBRL
Dimensions uses, then what model are you using?

If people are working with XBRL and they cannot answer that question, it seems
to me that they have a fundamental problem.

Relational model, hierarchical model, multidimensional model...computers need
some model to work with anything. When I was working on the US GAAP Taxonomy, I
got the sense that many others were looking at things as David vun Kannon
described "facts floating in space" on another thread. If you have individual
things floating around in space you still need some model, but it can be much
more less of a model. But, the US GAAP Taxonomy as boatloads of relations.
Things are simply not floating around in space.

A second point which goes along the lines of what Roland said about "true
dimensions" and "presentation related dimensions", etc. I have run across a
number of different schemes for working with information and I am seeing
consistencies between what Roland described in terms begin distinct about the
role of a dimension. I have seen the term "display" attributes used and
"navigation attributes" and "formatting information", etc. Seems to be very
consistent with what Roland is saying.

This little 90 second video articulates a model which this company uses and how
they separate their model (see point above about having SOME model) into
distinct pieces:

http://www.a3solutions.com/media/A3_SpreadsheetAutomation_preso.html
It seems to me to show very clearly that these folks separated "presentation"
(they use formatting) and the other stuff.

I suspect that a lot of people trying to model XBRL taxonomies are using some
sort of presentation model as their model. Not good.

So, what model are you using?

Cheers,

 Charlie