Machine learning based analysis of electronic communications

Patent metadata

Field	Value
Publication number	US-12190214-B2
Application number	US-202318376999-A
Country	US
Kind code	B2
Filing date	Oct 5, 2023
Priority date	Aug 12, 2020
Publication date	Jan 7, 2025
Grant date	Jan 7, 2025

Abstract

Aspects of the disclosure relate to machine learning based analysis of electronic communications. A computing platform may monitor receipt of a potentially unacceptable electronic communication by a user. Then, the computing platform may extract one or more attributes of the potentially unacceptable electronic communication. The computing platform may then perform, based on the one or more attributes, textual analysis of the potentially unacceptable electronic communication. Subsequently, the computing platform may retrieve one or more rules applicable to the potentially unacceptable electronic communication. Then, the computing platform may determine, based on the textual analysis and the one or more rules, and based on a repository of previously identified unacceptable content, whether the potentially unacceptable electronic communication is unacceptable. Subsequently, the computing platform may trigger, based on a determination that the potentially unacceptable electronic communication is unacceptable, one or more actions associated with the unacceptable electronic communication.

First claim

What is claimed is:

1. A computing platform, comprising:
at least one processor;
a communication interface communicatively coupled to the at least one processor; and
memory storing computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
extract one or more attributes of a potentially unacceptable electronic communication, the one or more attributes of the potentially unacceptable electronic communication including at least network information associated with transmission of the potentially unacceptable electronic communication and a list of computer network nodes that routed the potentially unacceptable electronic communication;
perform, based on the one or more attributes, textual analysis of the potentially unacceptable electronic communication;
retrieve one or more rules applicable to the potentially unacceptable electronic communication, the one or more rules including at least one rule based on a string generated from the one or more attributes of the potentially unacceptable electronic communication, wherein the at least one rule based on the string generated from the one or more attributes of the potentially unacceptable electronic communication includes classifying the potentially unacceptable electronic communication based on the string and wherein the string is a case-sensitive string;
determine, based on the textual analysis and the one or more rules, and based on a repository of previously identified unacceptable content, whether the potentially unacceptable electronic communication is unacceptable; and
trigger, based on a determination that the potentially unacceptable electronic communication is unacceptable, one or more actions associated with the unacceptable electronic communication.

2. The computing platform of claim 1 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
train a machine learning model to determine whether the potentially unacceptable electronic communication is unacceptable.

3. The computing platform of claim 2 , wherein the machine learning model comprises:
a first logistic regression model to determine whether the potentially unacceptable electronic communication is authorized;
a second logistic regression model to determine whether the potentially unacceptable electronic communication is spam;
a third logistic regression model to determine whether the potentially unacceptable electronic communication is a test communication; and
a fourth logistic regression model to determine whether the potentially unacceptable electronic communication is malicious.

4. The computing platform of claim 3 , wherein the machine learning model comprises, for each logistic regression model, a cost function, and wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
optimize the cost function.

5. The computing platform of claim 4 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
dynamically adjust, for each logistic regression model, a threshold associated with the cost function, wherein a threshold associated with the fourth logistic regression model is lower than a threshold associated with the first logistic regression model, the second logistic regression model and the third logistic regression model.

6. The computing platform of claim 3 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
aggregate an output of, the first logistic regression model, the second logistic regression model, the third logistic regression model, and the fourth logistic regression model, to determine whether the potentially unacceptable electronic communication is unacceptable.

7. The computing platform of claim 2 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
generate, based on an output of one or more regression models, labeled data to train the machine learning model.

8. The computing platform of claim 1 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
retrieve, from an electronic mailbox of a user, the potentially unacceptable electronic communication received by the user; and
extract the one or more attributes based on the retrieved potentially unacceptable electronic communication.

9. The computing platform of claim 1 , wherein the instructions comprise additional computer-readable instructions that, when executed by the at least one processor, cause the computing platform to:
perform the textual analysis based on natural language processing.

10. The computing platform of claim 1 , wherein the one or more attributes further comprise a subject, a header, a sender, a recipient, a body, a mail server, and a message identifier.

11. A method, comprising:
at a computing platform comprising at least one processor, a communication interface, and memory:
extracting one or more attributes of a potentially unacceptable electronic communication, the one or more attributes of the potentially unacceptable electronic communication including at least network information associated with transmission of the potentially unacceptable electronic communication and a list of computer network nodes that routed the potentially unacceptable electronic communication;
performing, based on the one or more attributes, textual analysis of the potentially unacceptable electronic communication;
retrieving one or more rules applicable to the potentially unacceptable electronic communication, the one or more rules including at least one rule based on a string generated from the one or more attributes of the potentially unacceptable electronic communication, wherein the at least one rule based on the string generated from the one or more attributes of the potentially unacceptable electronic communication includes classifying the potentially unacceptable electronic communication based on the string and wherein the string is a case-sensitive string;
determining, based on the textual analysis and the one or more rules, and based on a repository of previously identified unacceptable content, whether the potentially unacceptable electronic communication is unacceptable; and
triggering, based on a determination that the potentially unacceptable electronic communication is unacceptable, one or more actions associated with the unacceptable electronic communication.

12. The method of claim 11 , further comprising:
training a machine learning model to determine whether the potentially unacceptable electronic communication is unacceptable.

13. The method of claim 12 , further comprising:
applying a first logistic regression model to determine whether the potentially unacceptable electronic communication is authorized;
applying a second logistic regression model to determine whether the potentially unacceptable electronic communication is spam;
applying a third logistic regression model to determine whether the potentially unacceptable electronic communication is a test communication; and
applying a fourth logistic regression model to determine whether the potentially unacceptable electronic communication is malicious.

14. The metho

Assignees

• Bank Of America

Inventors

• Jacobson Daniel M
• Franklin-Barr Belinda
• Lewis Patrick

Classifications

• H04L51/212

  using filtering or selective blocking · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F40/263

  Language identification · CPC title

• H04L67/535 Primary

  Tracking the activity of the user (network monitoring arrangements H04L43/00; recording of computer activity G06F11/34) · CPC title