ThinkTWENTY20

There’s more to include in our analyses these days than the numbers on the report. With increasing structured legal requirements, Inline XBRL meshing visual reporting with invisible XBRL metadata, AI models and the interpretations, pulling together all of the pieces requires a new way of thinking. 

As part of the United Nations team I’ve been facilitating, comparing, mapping and leveraging different accounting and audit data standards, I’ve been performing substantial “data” analytics on the different models. I’ve been slicing and dicing data in the spreadsheets and schemas that are meant to better represent accounting data. I’ve been looking for tools to create “XPath expressions”, an unambiguous way to navigate and unambiguously reference the “data points”, as well as considering the need for a “language” (a textual convention) to convey the conditions, interrelationships, and indications of how precise the mapping of data ideas between the standards’ components are.

Data analytics ...what's old became new again. Data analytics is recognized by AACSB (an American professional organization that accredits business academic programs) as one of three primary components of current and emerging accounting and business practices where accounting degree programs need to include learning experiences to develop skills and knowledge for integrating information technology in accounting and business. (See standard A5 of the "Standards for Accounting Accreditation").

What I don’t see a lot of is data analytics that goes “up” to the requirements and “down” to the analyses; it all focused on the “obvious” data, the numbers in the reports. However, the bigger picture is (like the “turtles”): it’s “data all the way down”.

I’ve mentioned that a colleague of mine, Michel Biezunski, is working with the United Nations team I’ve led on comparing, mapping and leveraging different accounting and audit data standards, such as OECD SAF-T/SAF-P, UN/CEFACT Accounting and Audit artefacts and XBRL’s Global Ledger Taxonomy Framework. As an XBRL person, I’ve had it drummed into me to have a firm separation between data (“the reporting information”), the metadata (the “angle brackets” within the reporting document), and “taxonomy” (the meta-metadata). We skip a meta level there, as an Inline XBRL best exemplifies: you have the visible “data”; the contextual, behind the scenes, but still tied to the instance meta-data.

I don’t know if we are going to be able to apply “traditional” metamodeling theory to the business reporting supply chain. I have published my “LIBRARIES” model before setting up a closed loop business reporting environment. (Interested folks can follow my 2012 Prezi, Is XBRL a Proxy for a Complete Organizational Compliance Lifecycle at https://prezi.com/bdomo30plabt/is-xbrl-a-proxy-for-a-complete-organizational-compliance-lifecycle/). But some of the pieces … the “RIES” part – the research and interpretation and evaluation and synthesis … some fresh thought on the implications of Artificial Intelligence/Machine Learning inputs, outputs and processes … some good thinking in the PhD thesis by Michael Anderson last year, It’s Data All the Way Down: Exploring the Relationship Between Machine Learning and Data Management. (https://deepblue.lib.umich.edu/handle/2027.42/153458)