Digital Financial Reporting

In his paper, Future Directions for Semantic Systems, John Sowa points out the following:

No one can predict what innovations will be discovered in the future, but commercially successful systems must satisfy two criteria: first, they must solve problems for which a large number of people need solutions; second, they must have automated and semi-automated methods for acquiring, analyzing, and organizing the required knowledge.

Hamed Mousavi and I created a working proof-of-concept that is a semantic system.  We call the software application Pesseract. You can download the software (requires Windows), fire it up, and give it a spin.  We created a bunch of demo scripts that you can walk through.

If you cannot or don't want to download the application, you can watch a set of videos that show the semantic system in action.

To get the semantic system to work, we had to create the "semantics".  Those semantics are in the form of information related to several financial reporting schemes represented in XBRL.  There is at least one commercially available software application that also uses those semantics, XBRL Cloud. There is at least one open source software application that uses these same ideas, XBRLQuery.com.

We have taken information from these three and other implementations to create a high-level logical conceptualization of a business report. Keep in mind that a financial report is a type of business report.  There have been several iterations of this logical conceptualization.  OMG's Standard Business Report Model (SBRM) is the most current version of that logical conceptualization.

All of this is proven to work.  I can also explain how it works. The document Special Theory of Machine-based Automated Communication of Semantic Information of Financial Statements explains how it all works.

The Digital Financial Reporting Manifesto explains why all this is useful.  While all of my metadata works and allows for these ideas to be explored and tested, the metadata is not sufficient to be a complete financial reporting scheme.  A lot more metadata is necessary to provide a complete system.

Whether there is enough existing metadata to begin enabling machine learning based approaches to help create new metadata is unclear.  I don't think that there is quite enough yet to semi-automated metadata creation.  But, I could be wrong.

The next step is to turn the working proof-of-concept into one or more products.