Granted Patent B2

US 12,585,992 · App. 17/894,050 · Granted Mar 24, 2026

Inventors: David Gorham (McKinney, TX); Michael Don Wigley (Dallas, TX); Mark Stephen DeMichele (Ann Arbor, MI)

Assignee: ZixCorp Systems, Inc.

G06N20/00G06F40/279G06F40/30

Abstract

Claims (77)

1 . A method, comprising:

receiving a plurality of electronic messages from a database;

reading from each of the plurality of electronic messages an express indication indicating whether or not a classification applies to an electronic message of the plurality of electronic messages;

identifying a plurality of attributes included in the plurality of electronic messages;

determining a confidence level based on the plurality of attributes, wherein the confidence level is used by a machine learning engine to determine whether the classification applies to the electronic message;

from among the plurality of attributes, creating a first training set of attributes to which the classification is known to apply;

from among the plurality of attributes, creating a second set of attributes to which the classification is known not to apply; and

training the machine learning engine using the first and the second training sets;

receiving a first electronic message;

determining the first electronic message includes an express indication that either indicates the classification applies to the first electronic message or does not apply to the first electronic message, wherein the express indication comprises a word or a flag configured by a user;

identifying a first attribute of the first electronic message;

applying the machine learning engine to the first attribute to determine the classification applies to the first electronic message;

providing the first electronic message to an enforcer adapted to apply the classification to the first electronic message, wherein the enforcer applies the classification by performing one or more actions associated with the classification;

receiving a second electronic message;

identifying a second attribute of the second electronic message;

applying the machine learning engine to the second attribute to determine the classification does not apply to the second electronic message; and

bypassing the enforcer based on the determination.

2 . The method of claim 1 , wherein the application of the classification to the first electronic message is based on a regulatory policy.

3 . The method of claim 2 , wherein the application of the classification to the first electronic message comprises performing an action specified by the regulatory policy.

4 . The method of claim 3 , further comprising determining whether the regulatory policy applies to the first electronic message or the user.

5 . The method of claim 4 , wherein the machine learning engine is trained based on information indicating whether the regulatory policy applies to the first electronic message of the user.

6 . The method of claim 1 , wherein the express indication is in content of the first electronic message.

7 . The method of claim 1 , further comprising:

setting a value associated with an attribute of the machine learning engine to a first value when a corresponding message attribute has been identified in the first electronic message; and

setting the attribute of the machine learning engine to a second value when a corresponding message attribute is not identified in the second electronic message.

8 . A system, comprising:

a processor; and

a non-transitory computer readable medium, comprising instructions for:

receiving a plurality of electronic messages from a database;

reading from each of the plurality of electronic messages an express indication indicating whether or not a classification applies to an electronic message of the plurality of electronic messages;

identifying a plurality of attributes included in the plurality of electronic messages;

determining a confidence level based on the plurality of attributes, wherein the confidence level is used by a machine learning engine to determine whether the classification applies to the electronic message;

from among the plurality of attributes, creating a first training set of attributes to which the classification is known to apply;

from among the plurality of attributes, creating a second set of attributes to which the classification is known not to apply; and

training the machine learning engine using the first and the second training sets;

receiving a first electronic message;

determining the first electronic message includes an express indication that either indicates the classification applies to the first electronic message or does not apply to the first electronic message, wherein the express indication comprises a word or a flag configured by a user;

identifying a first attribute of the first electronic message;

applying the machine learning engine to the first attribute to determine the classification applies to the first electronic message;

providing the first electronic message to an enforcer adapted to apply the classification to the first electronic message, wherein the enforcer applies the classification by performing one or more actions associated with the classification;

receiving a second electronic message;

identifying a second attribute of the second electronic message;

applying the machine learning engine to the second attribute to determine the classification does not apply to the second electronic message; and

bypassing the enforcer based on the determination.

9 . The system of claim 8 , wherein the application of the classification to the first electronic message is based on a regulatory policy.

10 . The system of claim 9 , wherein the application of the classification to the first electronic message comprises performing an action specified by the regulatory policy.

11 . The system of claim 10 , wherein the instructions further comprise instructions for determining whether the regulatory policy applies to the first electronic message or the user.

12 . The system of claim 11 , wherein the machine learning engine is trained based on information indicating whether the regulatory policy applies to the first electronic message or the user.

13 . The system of claim 8 , wherein the express indication is in content of the first electronic message.

14 . The system of claim 8 , further comprising:

setting a value associated with an attribute of the machine learning engine to a first value when a corresponding message attribute has been identified in the first electronic message; and

setting the attribute the machine learning engine to a second value when a corresponding message attribute is not identified in the second electronic message.

15 . A non-transitory computer readable medium, comprising instructions for:

receiving a plurality of electronic messages from a database;

reading from each of the plurality of electronic messages an express indication indicating whether or not a classification applies to an electronic message of the plurality of electronic messages;

identifying a plurality of attributes included in the plurality of electronic messages;

determining a confidence level based on the plurality of attributes, wherein the confidence level is used by a machine learning engine to determine whether the classification applies to the electronic message;

from among the plurality of attributes, creating a first training set of attributes to which the classification is known to apply;

from among the plurality of attributes, creating a second set of attributes to which the classification is known not to apply; and

training the machine learning engine using the first and the second training sets;

receiving a first electronic message;

determining the first electronic message includes an express indication that either indicates the classification applies to the first electronic message or does not apply to the first electronic message, wherein the express indication comprises a word or a flag configured by a user;

identifying a first attribute of the first electronic message;

applying the machine learning engine to the first attribute to determine the classification applies to the first electronic message;

providing the first electronic message to an enforcer adapted to apply the classification to the first electronic message, wherein the enforcer applies the classification by performing one or more actions associated with the classification;

receiving a second electronic message;

identifying a second attribute of the second electronic message;

applying the machine learning engine to the second attribute to determine the classification does not apply to the second electronic message; and

bypassing the enforcer based on the determination.

16 . The non-transitory computer readable medium of claim 15 , wherein the application of the classification to the first electronic message is based on a regulatory policy.

17 . The non-transitory computer readable medium of claim 16 , wherein the application of the classification to the first electronic message comprises performing an action specified by the regulatory policy.

18 . The non-transitory computer readable medium of claim 17 , further comprising determining whether the regulatory policy applies to the first electronic message or the user.

19 . The non-transitory computer readable medium of claim 18 , wherein the machine learning engine is trained based on information indicating whether the regulatory policy applies to the first electronic message or the user.

20 . The non-transitory computer readable medium of claim 15 , wherein the express indication is in content of the first electronic message.

21 . The non-transitory computer readable medium of claim 15 , further comprising:

setting a value associated with an attribute of the machine learning engine to a first value when a corresponding message attribute has been identified in the first electronic message; and

setting the attribute the machine learning engine to a second value when a corresponding message attribute is not identified in the second electronic message.

Assignments (1)

Continuity (2)

References Cited (34)