Someone pointed me to a white paper as part of a discussion which occurred during the XBRL International Conference in Washington, D.C. last month. The white paper send me down a path which has become quite fruitful. I am still trying to synthesize what all this seems to mean, but here are the highlights.
The paper called "Getting Started with ADAPT" discusses a standardized OLAP database design. The reason the ADAPT model was created was that there is no OLAP standard design model. Different Business Intelligence (BI) applications who use OLAP implemented it somewhat differently. As someone put it, each of these applications "has an attitude" or a "personality", meaning that if you do something in one BI tool one way, it might look different in another BI tool. The ADAPT common model is used to minimize the effort and maximize the quality of BI applications consultants implement for business users.
The ADAPT model defines nine objects. These nine objects are: hypercube, dimension, hierarchy, level, member, attribute, scope, model, and context.
What is really interesting is that there is a high correlation between the ADAPT model and XBRLS (XBRL Simple Application Profile). The creators of XBRLS were totally unaware of the ADAPT model. Yet, there is a high correlation between XBRLS and ADAPT. I found that to be quite interesting.
But thinking about this more, I realized that of course they would be similar. They are trying to do the same thing. ADAPT and XBRLS are both used to model business information using a multidimensional model. (If you don't understand what OLAP or the multidimensional model is, this white paper does a good job of explaining these in terms a business user can understand.)
So what does this mean? Well, these are some of the things I seem to be seeing here and some questions which I have:
- More and more business information is making its way into Business Intelligence (BI) applications.
- XBRL is about modeling business information, which will likely find its way into BI applications at some point.
- XBRL (general XBRL, as specified by the XBRL Specification) is, well, too general to fit into BI applications easily. It can be done. Just like other data (this is not unique to XBRL), you can use all sorts of tricks and tools (often big heavy hammers) to cram poorly modeled data into whatever application. Another alternative is to start with a better information model.
- Is XBRL Simple Application Profile (XBRLS) a good approach to getting XBRL into BI applications? Seems to be.
- Or, is there a need for something like "XBRL Application Profile for OLAP"?
- Are BI tools the missing rendering engine needed by XBRL? I would speculate that the big BI vendors would really like that to be the case.
- Could XBRL become a way to achieve interoperabilty between BI applications, given that no BI/OLAP standard exists today?
Remember that business information is inherently multidimensional in nature; different entities, departments, periods, reporting scenarios, etc.