Granted Patent B1

US 12,645,641 · App. 18/666,095 · Granted Jun 2, 2026

Inventors: Ankit Goel (Oakton, VA); Annette Best (Ashburn, VA); Domenico Cacciavillani (River Vale, NJ); Deanna Jo Dabney (Stafford, VA); Deepthi Ganta (Broadlands, VA); Matthew Brian Maycott (Broadlands, VA); Keenan Moukarzel (Falls Church, VA); Samuel Edward Oliver, III (Aldie, VA); Thomas C. Schweikert (Leesburg, VA); Sukhdeep Kaur Sherry (Ashburn, VA)

Assignee: Federal Home Loan Mortgage Corporation (Freddie Mac)

G06F16/148G06F16/176G06F16/2365G06N20/00

Abstract

Claims (49)

1 . A system for recognizing data files in a file repository related to a transaction, the file repository containing a collection of data files generated by multiple separate entities and lacking consistent identifiers among related subsets of data files, the system comprising:

at least one non-transitory computer readable storage medium configured to store one or more machine learning algorithms, each machine learning algorithm trained to identify, based on a set of correlations between data file characteristics, multiple data files belonging to a shared transaction;

at least one machine-readable data store configured to store a plurality of data files, each data file of the plurality of data files corresponding to a respective transaction of a plurality of transactions, wherein

each respective subset of data files of a plurality of subsets of the plurality of data files was generated by a different respective system of a plurality of external systems, and

each file of the plurality of data files is associated with a respective submission link identifier of a plurality of submission link identifiers, each submission link identifier assigned to a different transaction of the plurality of transactions;

a submission linking service comprising processing circuitry configured to

using a set of identification attributes descriptive of a topic transaction, query the at least one machine-readable data store to identify a set of candidate data files of the plurality of data files, wherein

the set of identification attributes are derived from a submission data file, and

each file of the set of candidate data files comprises a corresponding attribute to at least one identification attribute of the set of identification attributes, and

prepare one or more model inputs for providing to at least one machine learning algorithm of the one or more machine learning algorithms, each model input of the one or more model inputs comprising, for each respective candidate data file of at least a portion of the set of candidate data files, i) a respective set of characteristic values, and ii) the respective submission link identifier of the respective candidate data file; and

a model service comprising processing circuitry configured to

obtain, from the submission linking service, the one or more model inputs,

using the one or more model inputs, perform machine learning analysis, by the one or more machine learning algorithms, to identify, from the set of candidate data files, each data file related to the topic transaction as represented by the set of identification attributes,

based on the machine learning analysis, determine an assigned submission link identifier for the submission data file, wherein to determine the assigned submission link identifier comprises

responsive to identifying one or more candidate data files sharing a same submission link identifier, assign, for the submission data file, the respective submission link identifier of the one or more candidate data files, and

responsive to failing to identify any related data files, assign, for the submission data file, a new submission link identifier added to the plurality of submission link identifiers, and

return, to the submission linking service, the assigned submission link identifier;

wherein the processing circuitry of the submission linking service is further configured to update the at least one machine-readable data store using the assigned submission link identifier.

2 . The system of claim 1 , wherein the set of identification attributes represents characteristics shared across a portion of the plurality of data files generated by two or more external systems of the plurality of external systems.

3 . The system of claim 1 , wherein the processing circuitry of the submission linking service is configured to:

receive the submission data file; and

derive, from the submission data file, the set of identification attributes.

4 . The system of claim 3 , wherein the submission linking service is configured to store, to the at least one machine-readable data store, the submission data file.

5 . The system of claim 1 , wherein to determine the assigned submission link identifier comprises, responsive to identifying two or more related data files having two or more different submission link identifiers, reassign at least one previously assigned link identifier of a first data file of the two or more related data files to the respective submission link identifier of another of the two or more related data files.

6 . The system of claim 5 , wherein the submission linking service is configured to store, to the at least one machine-readable data store, the reassigned link identifier in association with the first data file.

7 . The system of claim 1 , wherein the submission linking service is configured to provide, to a target external system of the plurality of external systems, the assigned submission link identifier.

8 . A method for recognizing one or more data files from a file repository that are related to a transaction, the file repository containing a collection of data files generated by multiple separate entities and lacking consistent identifiers among related subsets of data files, the method comprising:

identifying, by processing circuitry, a set of identification attributes descriptive of a topic transaction, wherein the set of identification attributes are derived from a submission data file;

using the set of identification attributes, querying, by the processing circuitry, at least one data store to identify a set of candidate data files of a plurality of data files, wherein

each data file of the plurality of data files i) corresponds to a respective transaction of a plurality of transactions, and ii) is associated with a respective submission link identifier of a plurality of submission link identifiers, each submission link identifier assigned to a different transaction of a plurality of transactions,

each subset of data files of a plurality of subsets of the plurality of data files originated from a different respective organizations of a plurality of third-party organizations, and

each file of the set of candidate data files comprises a corresponding attribute to at least one identification attribute of the set of identification attributes;

preparing, by the processing circuitry, one or more model inputs for providing to at least one machine learning algorithm of one or more machine learning algorithms, the one or more model inputs comprising, for each respective candidate data file of at least a portion of the set of candidate data files, a) a respective set of characteristic values, and b) the respective submission link identifier of the respective candidate data file;

using the at least one machine learning algorithm in coordination with the one or more model inputs, identifying, from the set of candidate data files, each data file related to the topic transaction as represented by the set of identification attributes, wherein

each machine learning algorithm is trained to identify, based on a set of correlations between data file characteristics, multiple data files belonging to a shared transaction;

determining, by the processing circuitry, topic transaction an assigned submission link identifier for the submission data file, wherein to determine the assigned submission link identifier comprises

responsive to identifying one or more candidate data files, assign, for the submission data file, the respective submission link identifier of the one or more candidate data files, and

responsive to failing to identify any related data files, assign, for the submission data file, a new submission link identifier added to the plurality of submission link identifiers; and

updating, by the processing circuitry, the at least one data store using the assigned submission link identifier.

9 . The method of claim 8 , wherein determining the assigned submission link identifier comprises assigning, to two or more related data files, a link identifier associated with the topic transaction.

10 . The method of claim 8 , wherein determining the assigned submission link identifier comprises identifying, within two or more related data files, a respective data file comprising an assigned link identifier.

11 . The method of claim 8 , wherein determining the assigned submission link identifier comprises reassigning at least one previously assigned link identifier of two or more additional data files to the assigned submission link identifier.

12 . The method of claim 8 , wherein the set of identification attributes represents characteristics shared across a portion of the plurality of data files originated from two or more third-party organizations of the plurality of third-party organizations.

13 . The method of claim 8 , further comprising receiving a query comprising the set of identification attributes.

14 . The method of claim 13 , wherein the query comprises a submission data file, the method further comprising deriving, by the processing circuitry from the submission data file, the set of identification attributes.

15 . The method of claim 14 , further comprising storing, to the at least one data store, the submission data file.

16 . The method of claim 8 , further comprising providing, to a remote system of a target third-party organization of the plurality of third-party organizations, the assigned submission link identifier.

17 . The method of claim 16 , further comprising receiving, from a second remote system, a query comprising the set of identification attributes, wherein

the second remote system is different than the remote system.

Continuity (2)

References Cited (42)