# US 12,010,075 B2 · Utilizing machine learning models to generate interactive digital text threads with personalized digital text reply options

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,010,075 B2
- Document ID: US12010075B2
- Status: Granted
- Kind: B2
- Application number: 17/809,765
- Filed: 2022-06-29
- Published: 2024-01-04
- Granted: 2024-06-11
- Earliest claimed: 2022-06-29
- Assignees: Chime Financial, Inc.
- Applicants: Chime Financial, Inc.
- Inventors: Jigar Mehta; Abbey Chaver; Abhi Sharma; Sashidhar Guntury
- Web page: https://patentgrep.com/patent/US12010075B2

## CPC

- H04L 51/02: using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages
- G06F 16/355: Creation or modification of classes or clusters
- H04L 51/046: Interoperability with other network applications or services
- H04L 51/216: Handling conversation history, e.g. grouping of messages in sessions or threads
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

The present disclosure relates to systems, non-transitory computer-readable media, and methods for utilizing a machine learning model to determine predicted client intent classifications and generate personalized digital text reply options within an automated interactive digital text thread. For example, disclosed systems utilize the machine learning model to generate predicted client intent classifications and corresponding intent classification probabilities. The disclosed systems utilize the predicted client disposition classifications and the disposition classification probabilities to generate personalized digital text reply options. Moreover, the disclosed systems can provide personalized digital text reply options to a client device within an automated interactive digital text thread, bypassing the inefficiency of menu options or protocols utilized to guide clients to terminal information.

## Claims

### Claim 1

A computer-implemented method comprising:
extracting client features corresponding to a client device participating in an automated interactive digital text thread;
identifying a hierarchical intent architecture comprising a plurality of intent classifications organized in a plurality of hierarchical layers;
generating, from the hierarchical intent architecture utilizing a machine learning model, a plurality of predicted client intent classifications and a plurality of intent classification probabilities from the client features;
applying a bonus weight to a first intent classification probability of the plurality of intent classification probabilities associated with a first predicted client intent classification from a first hierarchical layer of the plurality of hierarchical layers, the bonus weight based on the first hierarchical layer;
selecting the first predicted client intent classification and at least two one additional predicted client intent classification from the plurality of predicted client intent classifications utilizing the plurality of intent classification probabilities; and
providing, for display via the client device, at least two personalized digital text reply options corresponding to the first predicted client intent classification and the at least two one additional predicted client intent classification via the automated interactive digital text thread.

### Claim 2 (depends on 1)

The computer-implemented method of claim 1, further comprising: in response to user interaction with a personalized digital text reply option of the at least two personalized digital text reply options:
generating a digital text response corresponding to the personalized digital text reply option; and
adding the digital text response to the automated interactive digital text thread.

### Claim 3 (depends on 2)

The computer-implemented method of claim 2, further comprising: in response to user interaction with the personalized digital text reply option:
generating an additional plurality of digital text reply options corresponding to the digital text response; and
providing, for display, the additional plurality of digital text reply options via the automated interactive digital text thread.

### Claim 4 (depends on 1)

The computer-implemented method of claim 1, further comprising training the machine learning model by:
monitoring client interaction with the automated interactive digital text thread to determine a ground truth client intent; and
modifying parameters of the machine learning model by comparing the plurality of predicted client intent classifications and the ground truth client intent.

### Claim 5 (depends on 1)

The computer-implemented method of claim 1, wherein utilizing the machine learning model comprises generating the plurality of predicted client intent classifications and the plurality of intent classification probabilities utilizing one or more of a random forest model or gradient boosted decision tree model.

### Claim 6 (depends on 1)

The computer-implemented method of claim 1, further comprising:
extracting the client features by determining a previous intent from a previous interactive text thread corresponding to the client device; and
generating the plurality of predicted client intent classifications and the plurality of intent classification probabilities from the previous intent utilizing the machine learning model.

### Claim 7 (depends on 1)

The computer-implemented method of claim 1, further comprising:
applying an additional bonus weight to a second intent classification probability of the plurality of intent classification probabilities associated with a second predicted client intent classification from a second hierarchical layer of the plurality of hierarchical layers, the additional bonus weight based on the second hierarchical layer,
wherein selecting the at least one additional predicted client intent classification comprises selecting the second predicted client intent classification.

### Claim 8 (depends on 1)

The computer-implemented method of claim 1, wherein applying the bonus weight to the first intent classification probability comprises applying a weighted combination of the first intent classification probability and a weight based on a position of the first hierarchical layer within the hierarchical intent architecture.

### Claim 9

A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause a computer system to:
extract client features corresponding to a client device participating in an automated interactive digital text thread;
identify a hierarchical intent architecture comprising a plurality of intent classifications organized in a plurality of hierarchical layers;
generate, from the hierarchical intent architecture utilizing a machine learning model, a plurality of predicted client intent classifications and a plurality of intent classification probabilities from the client features;
apply a bonus weight to a first intent classification probability of the plurality of intent classification probabilities associated with a first predicted client intent classification from a first hierarchical layer of the plurality of hierarchical layers, the bonus weight based on the first hierarchical layer;
select the first predicted client intent classification and at least one additional predicted client intent classification from the plurality of predicted client intent classifications utilizing the plurality of intent classification probabilities; and
provide, for display via the client device, at least two personalized digital text reply options corresponding to the first predicted client intent classification and the at least one additional predicted client intent classification via the automated interactive digital text thread.

### Claim 10 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the instructions, when executed by the at least one processor, further cause the computer system to, in response to user interaction with a personalized digital text reply option of the at least two personalized digital text reply options:
generate a digital text response corresponding to the personalized digital text reply option; and
add the digital text response to the automated interactive digital text thread.

### Claim 11 (depends on 10)

The non-transitory computer-readable medium of claim 10, wherein the instructions, when executed by the at least one processor, further cause the computer system to, in response to user interaction with the personalized digital text reply option:
generate an additional plurality of digital text reply options corresponding to the digital text response; and
provide, for display, the additional plurality of digital text reply options via the automated interactive digital text thread.

### Claim 12 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the instructions, when executed by the at least one processor, further cause the computer system to train the machine learning model by:
monitoring client interaction with the automated interactive digital text thread to determine a ground truth client intent; and
modifying parameters of the machine learning model by comparing the plurality of predicted client intent classifications and the ground truth client intent.

### Claim 13 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the instructions, when executed by the at least one processor, further cause the computer system to: generate the plurality of predicted client intent classifications and the plurality of intent classification probabilities utilizing one or more of a random forest model or gradient boosted decision tree model.

### Claim 14 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the instructions, when executed by the at least one processor, further cause the computer system to:
identify a hierarchical intent architecture comprising a plurality of intent classifications organized in a plurality of hierarchical layers; and
select a first predicted client intent classification from a first hierarchical layer of the plurality of hierarchical layers and a second predicted client intent classification from a second hierarchical layer of the plurality of hierarchical layers.

### Claim 15

A system comprising:
at least one processor; and
at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:
extract client features corresponding to a client device participating in an automated interactive digital text thread;
identify a hierarchical intent architecture comprising a plurality of intent classifications organized in a plurality of hierarchical layers;
generate, from the hierarchical intent architecture utilizing a machine learning model, a plurality of predicted client intent classifications and a plurality of intent classification probabilities from the client features;
apply a bonus weight to a first intent classification probability of the plurality of intent classification probabilities associated with a first predicted client intent classification from a first hierarchical layer of the plurality of hierarchical layers, the bonus weight based on the first hierarchical layer;
select the first predicted client intent classification and at least one additional predicted client intent classification from the plurality of predicted client intent classifications utilizing the plurality of intent classification probabilities; and
provide, for display via the client device, at least two personalized digital text reply options corresponding to the first predicted client intent classification and the at least one additional predicted client intent classification via the automated interactive digital text thread.

### Claim 16 (depends on 15)

The system of claim 15, further comprising instructions that, when executed by the at least one processor, cause the system to in response to user interaction with a personalized digital text reply option of the at least two personalized digital text reply options:
generate a digital text response corresponding to the personalized digital text reply option; and
add the digital text response to the automated interactive digital text thread.

### Claim 17 (depends on 16)

The system of claim 16, further comprising instructions that, when executed by the at least one processor, cause the system to, in response to user interaction with the personalized digital text reply option:
generate an additional plurality of digital text reply options corresponding to the digital text response; and
provide, for display, the additional plurality of digital text reply options via the automated interactive digital text thread.

### Claim 18 (depends on 15)

The system of claim 15, further comprising instructions that, when executed by the at least one processor, cause the system to train the machine learning model by:
monitoring client interaction with the automated interactive digital text thread to determine a ground truth client intent; and
modifying parameters of the machine learning model by comparing the plurality of predicted client intent classifications and the ground truth client intent.

### Claim 19 (depends on 15)

The system of claim 15, further comprising instructions that, when executed by the at least one processor, cause the system to:
extract the client features by determining a previous intent from a previous interactive text thread corresponding to the client device; and
generate the plurality of predicted client intent classifications and the plurality of intent classification probabilities from the previous intent utilizing the machine learning model.

### Claim 20 (depends on 15)

The system of claim 15, further comprising instructions that, when executed by the at least one processor, cause the system to:
identify a hierarchical intent architecture comprising a plurality of intent classifications organized in a plurality of hierarchical layers; and
select a first predicted client intent classification from a first hierarchical layer of the plurality of hierarchical layers and a second predicted client intent classification from a second hierarchical layer of the plurality of hierarchical layers.