Digital Financial Reporting

I have been fiddling around with how to best use XBRL and have consolidated many, many, many other ideas into something that I am referring to as an "interactive information hypercube".

Breaking this down, this is what the notion of an interactive information hypercube is based on.  Again, these really are not my base ideas, I am just combining many other ideas together in order to achieve something which I believe needs to be achieved.  These ideas are:

• Interactive Information: The notion of interactive information comes from the term "interactive data" which was to the best of my knowledge coined by the US SEC.  I believe that "information" is a more appropriate term than "data" in the context in which I am working.
• Hypercube: Ever since I started trying to understand XBRL Dimensions, I never really understood the difference between a cube and a hypercube.  A couple of months ago I read something which clarified this, at least in my mind.  Everyone can probably visualize what a cube is.  A cube has three dimensions, it is a physical thing.  Some business data has three or less dimensions which can be made to fit into the three physical dimensions of a cube.  However, other business information has more than three dimensions which makes it difficult to visualize in the form of a cube.  A hypercube is something which can represent any number of dimensions.

Now, you really have to stretch your imagination a bit with this graphic.  But really take a look at the graphic. Imagine information expressed in that sort of form rather than on a piece of two dimensional paper!  That is the idea.  Clearly an application to view information would not look like that graphic; the point is that it does help one see the limitations of paper in communicating information.

This is a prototype "interactive information viewer" which I have been experimenting with during the process of creating XBRLS.  The prototype takes what I had referred to as "neutral format tables" in XBRLS, modifies the tables slightly, and organizes the "99-Combined" XBRLS meta pattern (which is really a combination of all the XBRLS meta patterns into one XBRL taxonomy and XBRL instance to test the patterns).  On the left, you can click on a hypercube from the XBRL taxonomy, and on the right a rendering of information relating to that hypercube is rendered in the form of a neutral format table.  The prototype is simply PNG images from an Excel spreadsheet.  The renderings were created manually in order to test the idea.  The next step is to automatically create the rendering from information in the XBRL taxonomy and XBRL instance. 

The prototype condenses down into an easier to work with set of hypercubes which you can view in this PDF.  A better example of the use case I am experimenting to try and make work with is a financial statement.  This PDF from the "comprehensive example"which I had created for XBRLS.  The larger example looks more like a financial statement and is therefore easier to relate to.  However, the XBRLS patterns in the 99-Combined example actually cover 100% of what is in the larger comprehensive example.  That is the point of the XBRLS meta patterns...that small set of meta patterns can be used to express literally anything which I have come across in either financial reporting or other areas of business reporting from my experience with such information.  Impossible you say?  Well, isn't it interesting that the fundamental concepts of addition, subtraction, multiplication and division in mathematics works in the domains of physics, business, chemistry, engineering, etc.  It is the simplicity of the meta patterns which offers the best evidence that they could be right.  Time and experimentation will tell.

An earlier version of this comprehensive example included the following experiment.  You can see the results of the experiment within this PDFwhich is similar to the PDF of the financial statement above enough to help you see the point I am about to make, but it is different (meaning, there is not a one to one correlation between the PDF files). The experiment was to express 100% of a financial statement within Excel pivot tables.  I did that and "printed" screen shots of the pivot tables organized within a Word document.

The point is this: A financial statement is a collection of hypercubes.

What I want to do is go back and redo the XBRLS comprehensive example using the same form as the prototype interactive information hypercube viewer from above.  That will be much easier for people to relate to and see that, in fact, (a) financial statements are collections of hypercubes and (b) that there are advantages to working with them as hypercubes, the primary benefit being that you can easily reorganize the financial information as you desire.

There are two things needed to make this work: an information model and a way to communicate flow.

XBRLS is the information model (at least one information model) which makes this work.  The COREP taxonomy will likely work this way also.

Flow is simply a mechanism for organizing the individual hypercubes in an order that you want.  That is actually easy to do, you can use an XBRL taxonomy to express flow.  I will go into that later.

There is another advantage to the notion of an interactive information hypercube that I can see.  Maybe I am right, maybe I am wrong.  Today, there is no "multidimensional model".  Each vendor implementing Business Intelligence (BI) software has their own model.  Similar, but different enough to make like more complicated than it needs to be for business users.  See "Getting Started with ADAPT".  This BI solutions provider Symmetry Corp outlined the issue, its ramifications, and their solution for it in that white paper.

What if one multidimensional model could be created which all software vendors used?  There is one SQL model.  Not perfect, but significantly more consistent between software vendors than the multidimensional model.  Who knows.

In a press release today, Oracle announced that their Enterprise Performance Management (EPM) System will support XBRL with help from UBmatrix.  An excerpt from the press release states:

To help publicly held companies facilitate the preparation, publishing and automatic exchange of financial statements in XBRL (eXtensible Business Reporting Language), Oracle will embed UBmatrix, Inc.’s leading XBRL technology into Oracle's Enterprise Performance Management (EPM) System, Oracle announced today.

You can read the press release yourself. Here is a link to information on what SAP is doing with XBRL. Not to leave out the company I work for, UBmatrix who is providing XBRL components for both SAP and Oracle, see here for more info.

In a related story, Investor's Business Daily says:

Today, most software companies — including UBmatrix, Germany's SAP (SAP), Microsoft (MSFT) and Oracle (ORCL) — are starting to create XBRL software to help companies send financial data over the Web.

I would probably add IBM to the list of companies who are creating XBRL software to help companies send financial (and other businessdata) over the Web.  IBM is pushing XBRL as a standard for risk reporting as you can see from this Web page and this Web page.  They are probably moving slowly but deliberately as they figure out how to best integrate XBRL into their offerings.

(If anyone knows specifically what IBM and Microsoft are doing, love to hear know that, maybe you could post a comment on this blog.)

The following is a graphic of an XBRL instance which, I believe, helps people visualize the connection between the "tags" and a "printed" business report be it on paper, in PDF or HTML, etc.

XBRL Instance GraphicOne could go on, and on, and on with this; connecting the concept to the XBRL taxonomy, the taxonomy concept to resources and relations. Wouldn't it be great if software applications did this for you!  Would help visualizing the information, finding errors, etc.  Clearly you would not want to see the tags, but rather interfaces that show the information contained within the tags. 

This link is to a high level model of XBRL.  The model was put together using the XBRL Specification and ideas from other models which I have seen.  I guess that this is more a physical model than a logical model.  I am no expert in creating models but I needed one and all the other models which I have come across were significantly lacking.  This model is certainly not a complete logical model which would need to go into significantly more detail, however it does contain a few logical notions within it such as the idea of networks, concepts, facts, and rules.

 High Level Model of XBRL

I think there are a number of different graphical models which are needed for XBRL.  A detailed physical model, a detailed logical model, etc.  If anyone is aware of these sorts of models, if you could make me aware of them that would be spectacular. Or, if anyone is aware of a better high level model than this one, please provide comments and I can either update this model or point to that better model.

Comments would be greatly appreciated.  This model is free for anyone to modify/improve, it is made available under a Creative Commons license.  I hope to create a graphical image of a logical model of the XBRLS application profile.