# US 2021/0241758 A1 · Automatically rectifying in real-time anomalies in natural language processing systems

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

- Number: US 2021/0241758 A1
- Document ID: US20210241758A1
- Status: Published application
- Kind: A1
- Application number: 16/779,280
- Filed: 2020-01-31
- Published: 2021-08-05
- Earliest claimed: 2020-01-31
- Assignees: Walmart Apollo, LLC
- Applicants: Walmart Apollo, LLC
- Inventors: Snehasish Mukherjee; Haoxuan Chen; Phani Ram Sayapaneni; Shankara Bhargava Subramanya
- Web page: https://patentgrep.com/patent/US20210241758A1

## CPC

- G06F 16/243: {Natural language query formulation}
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G06F 3/167: {Audio in a user interface, e.g. using voice commands for navigating, audio feedback}
- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G10L 15/30: Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G10L 2015/223: {Execution procedure of a spoken command}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

A method for automatically rectifying in real-time anomalies in natural language processing systems. The method can include receiving command data from a user device of a user. The command data can correspond to a user request. The method further can include retrieving, from a new template database, a new request template corresponding to the user request. Additionally, the method can include retrieving, from the new template database, an output instruction corresponding to the new request template, when the new request template is found. The method further can include determining, by a machine learning system, the output instruction corresponding to the user request, when the new request template is not found. The method also can include transmitting the output instruction to a request processing system. The request processing system can be configured to perform the output instruction and transmit, to the user device, a response to the user request. Other embodiments are disclosed.

## Claims

### Claim 1

A system comprising:
one or more processors; and
one or more non-transitory computer-readable media storing computing instructions configured to run on the one or more processors and perform:
receiving command data from a user device of a user, wherein the command data corresponds to a user request;
retrieving, from a new template database, a new request template corresponding to the user request;
when the new request template is found, retrieving, from the new template database, an output instruction corresponding to the new request template;
when the new request template is not found:
determining, by a machine learning subsystem, the output instruction corresponding to the user request; and
transmitting the output instruction to a request processing subsystem, wherein the request processing subsystem is configured to perform the output instruction and transmit, to the user device, a response to the user request.

### Claim 2 (depends on 1)

The system in claim 1, wherein:
the new template database comprises one or more new frequently-asked-questions (FAQ) templates and one or more corresponding new FAQ answers; and
when the one or more new FAQ templates comprise the new request template, the one or more corresponding new FAQ answers comprise the output instruction corresponding to the new request template.

### Claim 3 (depends on 1)

The system in claim 1, wherein:
the new template database comprises an in-memory database.

### Claim 4 (depends on 3)

The system in claim 3, wherein:
the computing instructions are further configured to perform:
loading at least a portion of the new template database at a frequency; and
the frequency is at least as great as a predetermined frequency threshold.

### Claim 5 (depends on 1)

The system in claim 1, wherein:
the command data comprise audio data; and
the computing instructions are further configured to perform:
determining, in real-time after receiving the command data, using automated speech recognition processing, the user request from the audio data.

### Claim 6 (depends on 5)

The system in claim 5, wherein:
the computing instructions are further configured to perform:
retrieving, from a new pattern database, in real-time after determining the user request from the audio data, a new pattern template corresponding to the user request; and
when the new pattern template is found, revising the user request based on the new pattern template.

### Claim 7 (depends on 6)

The system in claim 6, wherein:
the new pattern database comprises an in-memory database.

### Claim 8 (depends on 1)

The system in claim 1, wherein:
the computing instructions are further configured to perform:
extracting, prior to transmitting the output instruction to the user device, entity information from entity data of the user request.

### Claim 9 (depends on 8)

The system in claim 8, wherein:
the computing instructions are further configured to perform:
retrieving, from an entity rule database, one or more entity rules corresponding to the entity data; and
when the one or more entity rules are found, overwriting the entity information corresponding to the one or more entity rules.

### Claim 10 (depends on 9)

The system in claim 9, wherein:
the entity rule database comprises an in-memory database.

### Claim 11

A method being implemented via execution of computing instructions configured to run at one or more processors and stored at one or more non-transitory computer-readable media, the method comprising:
receiving command data from a user device of a user, wherein the command data corresponds to a user request;
retrieving, from a new template database, a new request template corresponding to the user request;
when the new request template is found, retrieving, from the new template database, an output instruction corresponding to the new request template;
when the new request template is not found:
determining, by a machine learning system, the output instruction corresponding to the user request; and
transmitting the output instruction to a request processing system, wherein the request processing system is configured to perform the output instruction and transmit, to the user device, a response to the user request.

### Claim 12 (depends on 11)

The method in claim 11, wherein:
the new template database comprises one or more new frequently-asked-questions (FAQ) templates and one or more corresponding new FAQ answers; and
when the one or more new FAQ templates comprise the new request template, the one or more corresponding new FAQ answers comprise the output instruction corresponding to the new request template.

### Claim 13 (depends on 11)

The method in claim 11, wherein:
the new template database comprises an in-memory database.

### Claim 14 (depends on 13)

The method in claim 13 further comprising:
loading at least a portion of the new template database at a frequency,
wherein:
the frequency is at least as great as a predetermined frequency threshold.

### Claim 15 (depends on 11)

The method in claim 11 further comprising:
determining, in real-time after receiving the command data, using automated speech recognition processing, the user request from audio data of the command data.

### Claim 16 (depends on 15)

The method in claim 15 further comprising:
retrieving, from a new pattern database, in real-time after determining the user request from the audio data, a new pattern template corresponding to the user request; and
when the new pattern template is found, revising the user request based on the new pattern template.

### Claim 17 (depends on 16)

The method in claim 16, wherein:
the new pattern database comprises an in-memory database.

### Claim 18 (depends on 11)

The method in claim 11 further comprising:
extracting, prior to transmitting the output instruction to the user device, entity information from entity data of the user request.

### Claim 19 (depends on 18)

The method in claim 18 further comprising:
retrieving, from an entity rule database, one or more entity rules corresponding to the entity data; and
when the one or more entity rules are found, overwriting the entity information corresponding to the one or more entity rules.

### Claim 20 (depends on 19)

The method in claim 19, wherein:
the entity rule database comprises an in-memory database.
