System and method using multiple profiles and scores for assessing financial transaction risk

Patent metadata

Field	Value
Publication number	US-12555115-B2
Application number	US-202318354508-A
Country	US
Kind code	B2
Filing date	Jul 18, 2023
Priority date	Dec 23, 2016
Publication date	Feb 17, 2026
Grant date	Feb 17, 2026

Abstract

In response to a request for risk assessment of a person-to-person (P2P) payment transaction, a risk assessment system returns a payer risk score, a payee risk score and a joint/fusion risk score. Risk scores are based on large amounts of data (including transaction data) provided by multiple financial institutions. The data is sorted and linked to individual people (e.g., common account holders), and assembled by a profile system into payee, payer and joint profiles for purposes of risk evaluation. A database associated with a profile system separately stores the profiles (and associated risk scores). Risk scores (payer, payee and joint) are provided for a transaction in response to a payer ID, a payee ID and transaction data.

First claim

What is claimed is:

1 . A risk management system, comprising:
one or more processors; and a memory having instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to:
receive, from an inquiring institution, a payer identifier of a payer of a transaction, a payee identifier of a payee of the transaction, and transaction data associated with the transaction;
retrieve a payer profile associated with the payer from a first discrete data structure and memory location, the payer profile comprising only data relevant to the payer profile;
retrieve a payee profile associated with the payee from a second discrete data structure and memory location, the payee profile comprising only data relevant to the payee profile;
receive transaction and account data associated with a plurality of past transactions of a general population of users;
perform, by a machine learning model, predictive analytics on the transaction and account data associated with the plurality of past transactions of the general population of users;
identify, by the machine learning model, characteristics of transactions that are relevant to risk of fraud based on a result of the predictive analytics;
develop risk assessment rules that indicate a probability of a given transaction being fraudulent based on the characteristics of transactions that are relevant to risk of fraud;
generate a payee score for the transaction based on application of the risk assessment rules to the payee profile and the transaction data, wherein generating the payee score comprises comparing characteristics of the transaction with past transactions of the payee from the payee profile;
generate a payer score for the transaction based on application of the risk assessment rules to the payer profile and the transaction data, wherein generating the payer score comprises comparing characteristics of the transaction with past transactions of the payer from the payer profile; and
send the payee score and the payer score to the inquiring institution.

2 . The risk management system of claim 1 , wherein:
the payer profile comprises a payer score and the payee profile comprises a payee score; and generating the payee score and the payer score comprises updating the payer score and the payee score.

3 . The risk management system of claim 1 , further comprising:
generating a joint score.

4 . The risk management system of claim 3 , wherein:
a joint profile associated with the payer and the payee is received with the payee profile and the payer profile; the joint profile comprises a joint score; and generating the joint score comprises updating the joint score.

5 . The risk management system of claim 1 , wherein:
the risk assessment rules are based on prior transaction data; and the prior transaction data comprises transactions identified as having a fraud risk.

6 . The risk management system of claim 1 , wherein:
the risk assessment rules are generated by a machine learning model that has analyzed characteristics of past transactions to identity characterizes of transactions that are relevant to fraud risk.

7 . A method of generating a risk score, comprising:
receiving, by a risk management system, a payer identifier of a payer of a transaction, a payee identifier of a payee of the transaction, and transaction data associated with the transaction from an inquiring institution; retrieving, by the risk management system, a payer profile associated with the payer from a first discrete data structure and memory location, the payer profile comprising only data relevant to the payer profile; retrieving a payee profile associated with the payee from a second discrete data structure and memory location, the payee profile comprising only data relevant to the payee profile; receiving, by the risk management system, transaction and account data associated with a plurality of past transactions of a general population of users; performing, by a machine learning model of the risk management system, predictive analytics on the transaction and account data associated with the plurality of past transactions of the general population of users; identifying characteristics of transactions that are relevant to risk of fraud based on a result of the predictive analytics; developing, by the risk management system, risk assessment rules that indicate a probability of a given transaction being fraudulent based on the characteristics of transactions that are relevant to risk of fraud; generating, by the risk management system, a payee score for the transaction based on application of the risk assessment rules to the payee profile and the transaction data, wherein generating the payee score comprises comparing characteristics of the transaction with past transactions of the payee from the payee profile; generating a payer score for the transaction based on application of the risk assessment rules to the payer profile and the transaction data, wherein generating the payer score comprises comparing characteristics of the transaction with past transactions of the payer from the payer profile; and sending, by the risk management system, the payee score and the payer score to the inquiring institution.

8 . The method of generating a risk score of claim 7 , further comprising:
normalizing the payee score and the payer score prior to sending the payee score and the payer score to the inquiring institution.

9 . The method of generating a risk score of claim 7 , further comprising:
generating, by the risk management system, the risk assessment rules using a machine learning model that has analyzed characteristics of past transactions to identity characterizes of transactions that are relevant to fraud risk.

10 . The method of generating a risk score of claim 9 , further comprising:
receiving, by the risk management system, additional transaction data; and updating the risk assessment rules based on the additional transaction data.

11 . The method of generating a risk score of claim 7 , wherein:
retrieving the payer profile is done using the payer identifier; and retrieving the payee profile is done using the payee identifier.

12 . The method of generating a risk score of claim 7 , wherein:
the payer identifier comprises one or both of a social security number of the payer and an account number of the payer; and the payee identifier comprises one or both of an email address of the payee and a telephone number of the payee.

13 . The method of generating a risk score of claim 7 , wherein:
retrieving the payer profile and the payee profile further comprises retrieving a joint profile associated with the payer and the payee; and the joint profile comprises transaction data from prior transactions between the payer and the payee.

14 . A non-transitory computer-readable medium having instructions stored thereon that, when executed by one or more processors, cause the one or more processors to:
receive, from an inquiring institution, a payer identifier of a payer of a transaction, a payee identifier of a payee of the transaction, and transaction data associated with the transaction; retrieve a payer profile associated with the payer from a first discrete data structure and memory location, the payer profile comprising only data relevant to the payer profile; retrieve a payee profile associated with the payee from a second discrete data structure and memory location, the payee profile comprising only data relevant to the payee profile; receive transaction and account data associated with a plurality of past trans

Assignees

• Early Warning Services Llc

Inventors

• Barnhardt David W
• Selway Anthony J

Classifications

• G06Q20/223

  based on the use of peer-to-peer networks · CPC title

• G06Q20/4016 Primary

  involving fraud or risk level assessment in transaction processing · CPC title