« Federal Energy Regulatory Commission (FERC) Adopting XBRL | Main | McKinsey: An Executive's Guide to AI »

XBRL Taxonomy Information Retrieval

Software vendors are providing a pretty poor level of functionality when it comes to finding information within an XBRL financial reporting taxonomy.  What is the state-of-the-art? Text search.

But I think there is a significantly better way, so I created this prototype to brainstorm and solidify my ideas.  I then created this prototype wizard. This is a web service where you provide an element name and it returns information for the requested report element name. This is the same information in machine-readable form.

But this is the best working prototype that I have for now.  That is for US GAAP, it does not incorporate the more advanced graph-type searching.  But, what this will allow me to do is communicate with software engineers and others better.

This prototype combines a bunch of things together and has a nicer user interface.

Not sure where to put these roll up relations. Ah!  I just had an idea.

The idea is that the search would leverage metadata and other information in order to provide a significantly more precise information retrial mechanism.

So the US GAAP XBRL Taxonomy has 17,077 elements.  That is a lot to work with.  But, you can break that full set into lots of different groupings.  For example, those 17,077 elements are comprised of:

  • 325 Tables
  • 251 Axis
  • 1,783 Members
  • 330 LineItems
  • 2,537 Abstracts
  • 183 Level 1 Note Text Blocks
  • 311 Level 2 Policy Text Blocks
  • 474 Level 3 Disclosure Text Blocks
  • 10,883 Concepts

Further, some of those groupings are still quite large so you can break them down by data type, period type. balance type, authoritative reference, etc.

If you did a raw text search on say "Cash" (you can try that here) you get 552 hits.  Somewhat helpful, but not quite what is needed.

But what if you could do a search and specify:

  • The concept contains the text string "Cash".
  • It relates to ASC topic "230"
  • You want only concepts
  • The concept is a "credit" and has a period type of "duration"
  • The concept is monetary in nature.
  • And it relates to the disclosure "CashFlowStatement"

I think you get the idea.  Lots of different ways to specify what you might be looking for.  Financial reporting concepts and terms, such as the US GAAP, are filled with a rich set of relations.  If you put that information into machine-readable terms, then the machines can perform what seems like magic.

Posted on Monday, June 17, 2019 at 10:04AM by Registered CommenterCharlie in | CommentsPost a Comment

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
All HTML will be escaped. Hyperlinks will be created for URLs automatically.