RedFile AI Publishes Open Source Classification Tree for Home Loan Documents

Open Source Mortgage Document Class Tree

Open Source Mortgage Document Class Tree

John Martin, CEO of RedFile AI LLC

John Martin, CEO

License-Free Classification Tree Is Step Towards Greater Consistency and Compatibility Across Lenders, Servicers, and Investors in Home Mortgage Industry

The open source RedFile classification tree avoids three common problems often associated with company-specific taxonomies…”

— John Martin

DALLAS, TEXAS, USA, January 20, 2021 / — John Martin, CEO of RedFile AI, announced today that RedFile has published an open-source document classification tree for documents found in home mortgage loan files, as developed over the course of processing millions of loan files. The document, available at, lists classifications that can be used for the documents commonly found in home mortgage loan files. Martin commented, “Common use of the open-source classifications can improve the profitability of loan purchases, sales, and servicing, and ultimately contribute to improved liquidity of the home mortgage loan market.”

Martin explained, “Consistent document classification is the foundation upon which most post-loan transactions are based, whether that’s selling or buying loans or servicing them. Without consistent classifications, you can’t, for example, be sure that there is actually a note to enforce in the event of foreclosure, and you can’t be sure that the servicer is charging the correct interest rates. Because virtually every servicer or investor uses their own classification scheme, every time a loan is resold or another servicer is used, the prior work done to validate the ‘tape’ data used to manage the loan must be repeated. There is a high cost of performing this redundant validation, particularly if the work is done manually by the loan analysts. At a company level these redundant validations make loan transactions less profitable, and at a macro level the high costs of validation make the home mortgage loan market less liquid.”

Martin noted, “The open source RedFile classification tree avoids three common problems often associated with company-specific taxonomies:

(1) Redundant Classifications, e.g., in some taxonomies, ‘IRS1099’ could be the same as ‘FedTax1099’ or ‘Tax-1099.’ These redundancies tend to creep when taxonomies have been used for years and those inconsistencies make it very difficult to automate downstream processes.

(2) Using Attribute Characteristics within Classifications. Some organizations mix document attributes with the basic classification, e.g., they might have ‘Loan Application’ as well as ‘Loan Application Executed,’ when it is more efficient to just have a “Loan Application” classification with a field or attribute to indicate that it was signed.

(3) Separate Use Case Classifications. Some companies have top-level classification categories for each major use case, e.g., Origination and Default, and include each document used for that use case as tier 2 classifications. This results in multiple, sometimes inconsistent classifications for what is the same document.”

Martin concluded, “Having a commonly-adopted, document classification scheme does not of course guarantee that each company classifies each document correctly, but it is a good start on achieving industry-wide uniformity.”

About RedFile AI. RedFile focuses on automated processing of home mortgage loan files by using unique 3DI algorithms and classification techniques to categorize and automate the manual document “stare and compare” prevalent in mortgage origination today. More information about RedFile AI is available at

John Martin
RedFile AI LLC
+1 972-687-9177
email us here

Source: EIN Presswire