US-20260127590-A1 - APPLICATION OF THE LAWS OF THERMODYNAMICS IN MONITORING THE DYNAMICS OF ACCOUNT-TO-ACCOUNT TRANSACTION SYSTEMS

US20260127590A1US 20260127590 A1US20260127590 A1US 20260127590A1US-20260127590-A1

Abstract

A system, computer-readable media and computer-implemented method for applying the laws of thermodynamics in monitoring the dynamics of account-to-account transaction systems. The computer-implemented method includes: receiving, from a data source, raw transaction records, each raw transaction record including a sender identifier (ID), recipient ID, an amount, and a fraud flag; extracting the sender ID, the recipient ID, and the fraud flag for each financial transaction; generating, based on the extracted information: sender transaction records; and recipient transaction; merging the sender transaction records and the recipient transaction records into potentially fraudulent transaction records and non-fraudulent transaction records; calculating a first transaction mass of the potentially fraudulent transaction records and a second transaction mass of the non-fraudulent transaction records; and calculating a first transaction velocity of the potentially fraudulent transaction records and a second transaction velocity of the non-fraudulent transaction records.

Inventors

Saeed Mirshekari
Mahdi JADALIHA

Assignees

MASTERCARD INTERNATIONAL INCORPORATED

Dates

Publication Date: 20260507
Application Date: 20241107

Claims (20)

1 . A non-transitory computer readable medium having instructions stored thereon that, when executed by at least one processor, cause the at least one processor to: receive, from a data source, raw transaction records corresponding to financial transactions, each of the raw transaction records including a sender identifier (ID), a recipient ID, a timestamp, an amount, and a fraud flag relating to a likelihood of fraud in the corresponding one of the financial transactions; extract, from the raw transaction records, the sender ID, the recipient ID, and the fraud flag for each of the financial transactions; generate, based on the sender ID, the recipient ID, and the fraud flag extracted for each of the raw transaction records: sender transaction records including non-fraudulent sender transaction records and potentially fraudulent sender transaction records; and recipient transaction records including non-fraudulent recipient transaction records and potentially fraudulent recipient transaction records; merge the sender transaction records and the recipient transaction records into potentially fraudulent transaction records and non-fraudulent transaction records; calculate a first transaction mass of the potentially fraudulent transaction records for each sender ID and recipient ID and a second transaction mass of the non-fraudulent transaction records for each sender ID and recipient ID; calculate a first transaction velocity of the potentially fraudulent transaction records and a second transaction velocity of the non-fraudulent transaction records; generate a visualization comprising dynamic metrics including any combination of two (2) or more of the first transaction mass, the second transaction mass, the first transaction velocity, or the second transaction velocity, said visualization including a graphical representation of at least one pattern associated with the potentially fraudulent transaction records and the non-fraudulent transaction records; analyze, with a machine learning model comprising a neural network, the visualization to identify a fraudulent pattern from the dynamic metrics; and render the visualization on a display.
2 . The non-transitory computer readable medium of claim 1 , the visualization including the first transaction mass and the second transaction mass.
3 . The non-transitory computer readable medium of claim 2 , wherein when executed by the at least one processor the computer-executable instructions further cause the at least one processor to: render a graphical user interface (GUI) on the display, said GUI including actionable items for modifying user-selectable parameters; receive a user input selecting one or more of the actionable items for setting the user-selectable parameters via the GUI; and generate the visualization based on said selections for the user-selectable parameters.
4 . The non-transitory computer readable medium of claim 3 , said user-selectable parameters including a time period of interest associated with a timestamp of each of the financial transactions, a total transaction mass, an average transaction mass, and an entity associated with each of the financial transactions.
5 . The non-transitory computer readable medium of claim 1 , the visualization including the first transaction velocity and the second transaction velocity.
6 . The non-transitory computer readable medium of claim 5 , wherein when executed by the at least one processor the computer-executable instructions further cause the at least one processor to: render a graphical user interface (GUI) on the display, said GUI including actionable items for modifying user-selectable parameters; receive a user input selecting one or more of the actionable items for setting the user-selectable parameters via the GUI; and generate the visualization based on said selections for the user-selectable parameters.
7 . The non-transitory computer readable medium of claim 6 , said user-selectable parameters including a time period of interest associated with a timestamp of each of the financial transactions, a total transaction mass, an average transaction mass, and an entity associated with each of the financial transactions.
8 . The non-transitory computer readable medium of claim 1 , wherein generating the sender transaction records and the recipient transaction records includes: aggregating those of the transaction records having matching sender IDs and matching corresponding ones of the fraud flags to generate the non-fraudulent sender transaction records and the potentially fraudulent sender transaction records; and aggregating those of the transaction records having matching recipient IDs and matching corresponding ones of the fraud flags to generate the non-fraudulent recipient transaction records and the potentially fraudulent recipient transaction records.
9 . The non-transitory computer readable medium of claim 1 , wherein merging the sender transaction records and the recipient transaction records includes: matching the sender IDs of the non-fraudulent sender transaction records and of the potentially fraudulent sender transaction records with the recipient IDs of the non-fraudulent recipient transaction records and of the potentially fraudulent recipient transaction records to generate an account ID for each match; merging the non-fraudulent sender transaction records and the non-fraudulent recipient transaction records having matched account IDs to generate the non-fraudulent transaction records; and merging the potentially fraudulent sender transaction records and the potentially fraudulent recipient transaction records having matched account IDs to generate the potentially fraudulent transaction records.
10 . The non-transitory computer readable medium of claim 1 , said first transaction mass including a sum of the amounts of the non-fraudulent transaction records and said second transaction mass including a sum of the amounts of the potentially fraudulent transaction records; said first transaction velocity including a transaction count of the non-fraudulent transaction records and said second transaction velocity including a transaction count of the potentially fraudulent transactions.
11 . A computer-implemented method, comprising: receiving, from a data source, raw transaction records corresponding to financial transactions, each raw transaction record including a sender identifier (ID), recipient ID, an amount, and a fraud flag relating to a likelihood of fraud in the corresponding one of the financial transactions; extracting, from the raw transaction records, the sender ID, the recipient ID, and the fraud flag for each of the financial transactions; generating, based on the sender ID, the recipient ID, and the fraud flag extracted for each of the raw transaction records: sender transaction records including non-fraudulent sender transaction records and potentially fraudulent sender transaction records; and recipient transaction records including non-fraudulent recipient transaction records and potentially fraudulent recipient transaction records; merging the sender transaction records and the recipient transaction records into potentially fraudulent transaction records and non-fraudulent transaction records; calculating a first transaction mass of the potentially fraudulent transaction records and a second transaction mass of the non-fraudulent transaction records; calculating a first transaction velocity of the potentially fraudulent transaction records and a second transaction velocity of the non-fraudulent transaction records; generating a visualization comprising dynamic metrics including any combination of two (2) or more of the first transaction mass, the second transaction mass, the first transaction velocity, or the second transaction velocity, said visualization including a graphical representation of at least one pattern associated with the potentially fraudulent transaction records and the non-fraudulent transaction records; analyzing, with a machine learning model comprising a neural network, the visualization to identify a fraudulent pattern from the dynamic metrics; and rendering the visualization on a display.
12 . The computer-implemented method of claim 11 , the visualization including the first transaction mass and the second transaction mass.
13 . The computer-implemented method of claim 12 , further comprising: rendering a graphical user interface (GUI) on the display, said GUI including actionable items for modifying user-selectable parameters; receiving a user input selecting one or more of the actionable items for setting the user-selectable parameters via the GUI; and generating the visualization based on said selections for the user-selectable parameters.
14 . The computer-implemented method of claim 13 , said user-selectable parameters including a time period of interest associated with a timestamp of each of the financial transactions, a total transaction mass, an average transaction mass, and an entity associated with each financial transaction.
15 . The computer-implemented method of claim 11 , the visualization including the first transaction velocity and the second transaction velocity.
16 . The computer-implemented method of claim 15 , further comprising: rendering a graphical user interface (GUI) on the display, said GUI including actionable items for modifying user-selectable parameters; receiving a user input selecting one or more of the actionable items for setting the user-selectable parameters via the GUI; and generating the visualization based on said selections for the user-selectable parameters.
17 . The computer-implemented method of claim 16 , said user-selectable parameters including a time period of interest associated with a timestamp of each of the financial transactions, a total transaction velocity, an average transaction velocity, and an entity associated with each of the financial transactions.
18 . The computer-implemented method of claim 11 , wherein generating the sender transaction records and the recipient transaction records includes: aggregating those of the transaction records having matching sender IDs and matching corresponding ones of the fraud flags to generate the non-fraudulent sender transaction records and the potentially fraudulent sender transaction records; and aggregating those of the transaction records having matching recipient IDs and matching corresponding ones of the fraud flags to generate the non-fraudulent recipient transaction records and the potentially fraudulent recipient transaction records.
19 . The computer-implemented method of claim 11 , wherein merging the sender transaction records and the recipient transaction records includes: matching the sender IDs of the non-fraudulent sender transaction records and of the potentially fraudulent sender transaction records with the recipient IDs of the non-fraudulent recipient transaction records and of the potentially fraudulent recipient transaction records to generate an account ID for each match; merging the non-fraudulent sender transaction records and the non-fraudulent recipient transaction records having matched account IDs to generate the non-fraudulent transaction records; and merging the potentially fraudulent sender transaction records and the potentially fraudulent recipient transaction records having matched account IDs to generate the potentially fraudulent transaction records.
20 . The computer-implemented method of claim 11 , said first transaction mass including a sum of the amounts of the non-fraudulent transaction records and said second transaction mass including a sum of the amounts of the potentially fraudulent transaction records; said first transaction velocity including a transaction count of the non-fraudulent transaction records and said second transaction velocity including a transaction count of the potentially fraudulent transaction records.

Description

FIELD OF THE INVENTION The present disclosure generally relates to systems, computer-readable media and computer-implemented methods for applying the laws of thermodynamics in monitoring the dynamics of account-to-account (A2A) transaction systems. BACKGROUND Monitoring patterns of behavior associated with fraudulent transactions typically involves computationally complex calculations that use a significant amount of computational and financial resources. For example, monitoring patterns of fraudulent behavior may involve memory intensive data transformations that consume significant amounts of electrical energy. Moreover, data used to monitor the patterns of fraudulent behavior may be days or weeks old, which can hinder the ability to detect fraud in real time. Moreover, account-to-account (A2A) transaction systems are particularly vulnerable to fraudulent transactions, as funds may be transferred directly from one account to another with minimal oversight. BRIEF SUMMARY In a first aspect, a system for applying the laws of thermodynamics in monitoring the dynamics of account-to-account transaction systems may be provided. The system may comprise a server. The server may include a communication element, a memory element, and a processing element which executes a software application. The software application may include instructions to: receive, from a data source, raw transaction records corresponding to financial transactions, each of the raw transaction records including a sender identifier (ID), a recipient ID, a timestamp, an amount, and a fraud flag relating to a likelihood of fraud in the corresponding one of the financial transactions; extract, from the raw transaction records, the sender ID, the recipient ID, and the fraud flag for each of the financial transactions; generate, based on the sender ID, the recipient ID, and the fraud flag extracted for each of the raw transaction records: sender transaction records including non-fraudulent sender transaction records and potentially fraudulent sender transaction records; and recipient transaction records including non-fraudulent recipient transaction records and potentially fraudulent recipient transaction records; merge the sender transaction records and the recipient transaction records into potentially fraudulent transaction records and non-fraudulent transaction records; calculate a first transaction mass of the potentially fraudulent transaction records for each sender ID and recipient ID and a second transaction mass of the non-fraudulent transaction records for each sender ID and recipient ID; and calculate a first transaction velocity of the fraudulent transaction records and a second transaction velocity of the non-fraudulent transaction records. The system may include additional, less, or alternate functionality, including that discussed elsewhere herein. In another aspect, a computer-implemented method for applying the laws of thermodynamics in monitoring the dynamics of account-to-account transaction systems may be provided. The computer-implemented method may include: receiving, from a data source, raw transaction records corresponding to financial transactions, each raw transaction record including a sender identifier (ID), recipient ID, an amount, and a fraud flag relating to a likelihood of fraud in the corresponding one of the financial transactions; extracting, from the raw transaction records, the sender ID, the recipient ID, and the fraud flag for each of the financial transactions; generating, based on the sender ID, the recipient ID, and the fraud flag extracted for each of the raw transaction records: sender transaction records including non-fraudulent sender transaction records and potentially fraudulent sender transaction records; and recipient transaction records including non-fraudulent recipient transaction records and potentially fraudulent recipient transaction records; merging the sender transaction records and the recipient transaction records into potentially fraudulent transaction records and non-fraudulent transaction records; calculating a first transaction mass of the potentially fraudulent transaction records and a second transaction mass of the non-fraudulent transaction records; and calculating a first transaction velocity of the potentially fraudulent transaction records and a second transaction velocity of the non-fraudulent transaction records. The method may include additional, less, or alternate actions, including those discussed elsewhere herein. In still another aspect, a system for applying the laws of thermodynamics in monitoring the dynamics of account-to-account transaction systems may be provided. The system may include a non-transitory computer-readable medium with a program stored thereon, wherein the program instructs one or more hardware processing elements to: receive, from a data source, raw transaction records corresponding to financial transactions, each of the raw transaction records includi