Granted Patent B2

US 12,307,396 · App. 17/814,413 · Granted May 20, 2025

Inventors: Hector Flores (Phoenix, AZ); Abhishek Jain (Phoenix, AZ); Robin Jain (Phoenix, AZ); Yogaraj Jayaprakasam (Phoenix, AZ); Srinivas K. Kumandan (Chandler, AZ); Jordan Meyerowitz (Phoenix, AZ)

Assignee: AMERICAN EXPRESS TRAVEL RELATED SERVICES COMPANY, INC.

G06Q10/063116G06F17/18G06Q10/063114G06Q10/06316

Abstract

Claims (59)

1. A method comprising:

training, by a processor, a long short-term memory network of a machine learning system using at least an event log received from a customer relationship management platform, training the long short-term memory network comprises generating a plurality of trained parameters for calculating a respective task win probability;

determining, by the processor, a task success for each of a plurality of previous tasks based at least in part on the event log;

generating, by the processor, a plurality of child user task schedules based at least in part on a first user task schedule from a plurality of user task schedules and a second user task schedule from the plurality of user task schedules, the first user task schedule and the second user task schedule each having a plurality of tasks and each task being assigned a respective time slot;

calculating, by the processor, a plurality of task win probabilities of the plurality of tasks at the respective time slot using the machine learning system, the machine learning system using the long short-term memory network with the plurality of trained parameters to determine the plurality of task win probabilities based at least in part on task metadata associated with the plurality of tasks;

determining, by the processor, a task progress for each of the plurality of tasks based at least in part on the task metadata;

calculating, by the processor, a plurality of total task win probabilities for the plurality of child user task schedules based at least in part on the plurality of task win probabilities, the task progress for each of the plurality of tasks, and the task success for each of the plurality of previous tasks;

determining, by the processor, an optimized user task schedule by selecting a respective user task schedule having a greatest total task win probability from a subset of the plurality of child user task schedules stored in a distributed cache;

transmitting, by the processor, the optimized user task schedule to the customer relationship management platform, the optimized user task schedule being accessible for display to a client device via a user interface of the customer relationship management platform; and

transmitting, by the processor, an alert notification to the client device associated with a user identifier in response to determining the optimized user task schedule, the alert notification indicating the determination of the optimized user task schedule at the customer relationship management platform.

2. The method of claim 1 , wherein the subset of the plurality of child user task schedules are stored in the distributed cache based at least in part on a first respective total task win probability of a first child user task schedule being greater than a second respective total task win probability of a second child user task schedule.

3. The method of claim 2 , wherein the subset of the plurality of child user task schedules is a first subset of the plurality of child user task schedules, and further comprising:

disregarding a second subset of the plurality of child user task schedules from the distributed cache.

4. The method of claim 1 , further comprising:

tracking, by the processor, a current task progress of a respective task using the machine learning system based at least in part on the event log received from the customer relationship management platform.

5. The method of claim 1 , wherein selecting the respective user task schedule having the greatest total task win probability further comprises:

polling from the subset of the plurality of child user task schedules stored in the distributed cache in order to compare a respective total task win probability for each of the plurality of child user task schedules.

6. The method of claim 1 , wherein the machine learning system retrieves the task metadata from a scheduling database based at least in part on a task identifier associated with a respective task.

7. The method of claim 1 , wherein calculating the plurality of task win probabilities of the plurality of tasks at the respective time slot using the machine learning system further comprises estimating a probability of closing a respective task with a win at the respective time slot.

8. A system comprising:

a computing device comprising a processor;

a distributed cache; and

machine-executable instructions stored in a memory that, in response to execution by the processor, cause the computing device to at least:

train a long short-term memory network of a machine learning system using at least an event log received from a customer relationship management platform;

determine a task success for each of a plurality of previous tasks based at least in part on the event log;

generate a plurality of child user task schedules based at least in part on a first user task schedule from a plurality of user task schedules and a second user task schedule from the plurality of user task schedules, the first user task schedule and the second user task schedule each having a plurality of tasks and each task being assigned a respective time slot;

calculate a plurality of task win probabilities of the plurality of tasks at the respective time slot using the machine learning system, the machine learning system using the long short-term memory network to determine the plurality of task win probabilities based at least in part on task metadata associated with the plurality of tasks;

determine a task progress for each of the plurality of tasks based at least in part on the task metadata;

calculate a plurality of total task win probabilities for the plurality of child user task schedules based at least in part on the plurality of task win probabilities, the task progress for each of the plurality of tasks, and the task success for each of the previous tasks;

determine an optimized user task schedule by selecting a respective user task schedule having a greatest total task win probability from a subset of the plurality of child user task schedules stored in the distributed cache;

transmit the optimized user task schedule to the customer relationship management platform, the optimized user task schedule being accessible for display to a client device via a user interface of the customer relationship management platform; and

transmit an alert notification to the client device associated with a user identifier in response to determining the optimized user task schedule, the alert notification indicating the determination of the optimized user task schedule at the customer relationship management platform.

9. The system of claim 8 , wherein the subset of the plurality of child user task schedules are stored in the distributed cache based at least in part on a first respective total task win probability of a first child user task schedule being greater than a second respective total task win probability of a second child user task schedule.

10. The system of claim 9 , wherein the subset of the plurality of child user task schedules is a first subset of the plurality of child user task schedules, and the machine-executable instructions stored in the memory, in response to execution by the processor, cause the computing device to at least:

delete a second subset of the plurality of child user task schedules from the distributed cache.

11. The system of claim 8 , wherein the machine-executable instructions stored in the memory, in response to execution by the processor, cause the computing device to at least:

track a current task progress of a respective task using the machine learning system based at least in part on the event log received from the customer relationship management platform.

12. The system of claim 8 , wherein selecting the respective user task schedule having the greatest total task win probability further causes the computing device to at least:

compare a respective total task win probability for each of the plurality of child user task schedules stored in the distributed cache.

13. The system of claim 8 , wherein the machine learning system retrieves the task metadata from a scheduling database based at least in part on a task identifier associated with a respective task.

14. The system of claim 8 , wherein calculating the plurality of task win probabilities of the plurality of tasks at the respective time slot using the machine learning system further comprises estimating a probability of closing a respective task with a win at the respective time slot.

15. A non-transitory, computer-readable medium comprising machine readable instructions that, when executed by a processor of a computing device, cause the computing device to at least:

train a long short-term memory network of a machine learning system using at least an event log received from a customer relationship management platform;

determine a task success for each of a plurality of previous tasks based at least in part on the event log;

generate a plurality of child user task schedules based at least in part on a first user task schedule from a plurality of user task schedules and a second user task schedule from the plurality of user task schedules, the first user task schedule and the second user task schedule each having a plurality of tasks and each task being assigned a respective time slot;

calculate a plurality of task win probabilities of the plurality of tasks at the respective time slot using the machine learning system, the machine learning system using the long short-term memory network to determine the plurality of task win probabilities based at least in part on task metadata associated with the plurality of tasks;

determine a task progress for each of the plurality of tasks based at least in part on the task metadata;

calculate a plurality of total task win probabilities for the plurality of child user task schedules based at least in part on the plurality of task win probabilities, the task progress for each of the plurality of tasks, and the task success for each of the previous tasks;

determine an optimized user task schedule by selecting a respective user task schedule having a greatest total task win probability from a subset of the plurality of child user task schedules stored in a distributed cache;

transmit the optimized user task schedule to the customer relationship management platform, the optimized user task schedule being accessible for display to a client device via a user interface of the customer relationship management platform; and

transmit an alert notification to the client device associated with a user identifier in response to determining the optimized user task schedule, the alert notification indicating the determination of the optimized user task schedule at the customer relationship management platform.

16. The non-transitory, computer-readable medium of claim 15 , wherein the subset of the plurality of child user task schedules are stored in the distributed cache based at least in part on a first respective total task win probability of a first child user task schedule being greater than a second respective total task win probability of a second child user task schedule.

17. The non-transitory, computer-readable medium of claim 15 , wherein the subset of the plurality of child user task schedules is a first subset of the plurality of child user task schedules, and the instructions, when executed by the processor of the computing device, cause the computing device to at least:

delete a second subset of the plurality of child user task schedules from the distributed cache.

18. The non-transitory, computer-readable medium of claim 15 , wherein the instructions, when executed by the processor of the computing device, cause the computing device to at least:

track a current task progress of a respective task using the machine learning system based at least in part on the event log received from the customer relationship management platform.

19. The non-transitory, computer-readable medium of claim 15 , wherein selecting the respective user task schedule having the greatest total task win probability further causes the computing device to at least:

compare a respective total task win probability for each of the plurality of child user task schedules stored in the distributed cache.

20. The non-transitory, computer-readable medium of claim 15 , wherein training the long short-term memory network comprises generating a plurality of trained parameters for calculating a respective task win probability.

Assignments (1)

Continuity (2)

References Cited (52)