# US 2021/0342693 A1 · Context and domain sensitive spelling correction in a database

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

- Number: US 2021/0342693 A1
- Document ID: US20210342693A1
- Status: Published application
- Kind: A1
- Application number: 17/379,065
- Filed: 2021-07-19
- Published: 2021-11-04
- Earliest claimed: 2017-08-18
- Assignees: MyFitnessPal, Inc.
- Applicants: Myfitnesspal, Inc.
- Inventors: Wenzhe Xue; Hesamoddin Salehian; Patrick Howell
- Web page: https://patentgrep.com/patent/US20210342693A1

## CPC

- G06N 3/08: Learning methods
- G16H 20/60: relating to nutrition control, e.g. diets
- G06F 16/35: Clustering; Classification
- G06F 16/353: into predefined classes
- G06F 16/355: Creation or modification of classes or clusters
- G06F 40/232: Orthographic correction, e.g. spell checking or vowelisation
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/35: Discourse or dialogue representation
- G06F 40/56: Natural language generation
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/09: Supervised learning
- G06N 20/00: Machine learning
- G16H 20/30: relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
- G16H 40/67: for remote operation
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method of operating a health tracking system is disclosed. The method comprises: receiving a first data record comprising at least a first descriptive string regarding a consumable item, the first descriptive string having at least one word thereof incorrectly spelled; generating a vector using the first descriptive string using a machine learning model; identifying a second descriptive string which corresponds to the consumable item and which has a correct spelling of the at least one incorrectly spelled word by applying the machine learning model to the generated vector; calculating a confidence factor regarding the identified second descriptive string using the machine learning model; and when it is determined that the confidence factor exceeds a predetermined threshold, (i) modifying the first data record by replacing the first descriptive string with the second descriptive string, and (ii) storing the modified first data record in the database.

## Claims

### Claim 1

A method of operating a health tracking system, the method comprising:
receiving a first data record comprising at least a first descriptive string and nutritional data regarding a consumable item, the first descriptive string having at least one word thereof having a first character sequence;
generating a data storage element with the first descriptive string of the first data record using a machine learning model;
identifying a second descriptive string which corresponds to the consumable item and which has a second character sequence for the at least one word by applying the machine learning model to the generated data storage element;
calculating a confidence factor regarding the identified second descriptive string using the machine learning model; and
when it is determined that the confidence factor exceeds a predetermined threshold:
modifying the first data record by replacing the first descriptive string with the second descriptive string; and
storing the modified first data record in a database.

### Claim 2 (depends on 1)

The method according to claim 1, wherein the act of generating the data storage element comprises generating a vector including:
generating a plurality of values corresponding to respective ones of a sequence of characters that comprises the at least one word of the first descriptive string; and
forming the vector using the generated plurality of values.

### Claim 3 (depends on 1)

The method according to claim 1, wherein the first character sequence is an incorrect spelling of the at least one word and the second character sequence is a correct spelling of the at least one word wherein the act of identifying the second descriptive string comprises:
determining, for each character of a sequence of characters that comprises the second descriptive string, a most probable correct character by applying the machine learning model to the generated data storage element; and
forming the second descriptive string using the determined most probable correct characters.

### Claim 4 (depends on 3)

The method according to claim 3, wherein the act of calculating the confidence factor comprises:
calculating the confidence factor based on probabilities of the determined most probable correct characters.

### Claim 5 (depends on 1)

The method according to claim 1, further comprising:
storing a plurality of data records in the database, each of the plurality of data records comprising at least a descriptive string and nutritional data regarding a consumable item;
identifying a subset of the plurality of data records; and
training the machine learning model using the identified subset of the plurality of data records.

### Claim 6 (depends on 5)

The method according to claim 5, wherein the first character sequence is an incorrect spelling of the at least one word and the second character sequence is a correct spelling of the at least one word wherein the act of identifying the subset of the plurality of data records comprises:
filtering the plurality of data records to identify data records in the plurality of data records in which the respective descriptive strings have correct spellings of every word contained therein.

### Claim 7 (depends on 5)

The method according to claim 5, wherein the act of training the machine learning model comprises:
generating, for each data record in the identified subset of the plurality of data records, a plurality of companion descriptive strings having at least one word thereof incorrectly spelled;
training the machine learning model using pairs of descriptive strings, each pair of descriptive strings including (i) the descriptive string of a respective data record in the identified subset of the plurality of data records and (ii) a corresponding one of the companion descriptive strings having at least one word thereof incorrectly spelled.

### Claim 8 (depends on 7)

The method according to claim 7, wherein the act of generating the plurality of companion descriptive strings comprises:
introducing one or more spelling errors into the descriptive strings of the identified subset of the plurality of data records.

### Claim 9 (depends on 1)

The method according to claim 1, wherein the machine learning model includes an encoder having at least one long short term memory layer and a decoder having at least one long short term memory layer.

### Claim 10 (depends on 1)

The method according to claim 1, further comprising:
prior to generating the data storage element, appending padding characters to the first descriptive string such that the first descriptive string has predetermined character length.

### Claim 11

A health tracking system comprising:
a database configured to store a plurality of data records, each of the plurality of data records comprising at least a descriptive string and nutritional data regarding a respective consumable item; and
a data processor in communication with the database, the data processor being configured to:
filter the plurality of data records to identify a subset thereof, the subset comprising those records in which the respective descriptive strings have a first type of spelling of every word contained therein;
generate, for each data record in the identified subset of the plurality of data records, a plurality of companion descriptive strings, each of the companion descriptive strings comprising one of a second type of spelling of at least one word contained therein;
train a machine learning model using pairs of descriptive strings, each pair of descriptive strings including (i) the descriptive string of a respective data record in the identified subset of the plurality of data records, and (ii) a corresponding one of the companion descriptive strings having at least one word thereof with the second type of spelling;
receive a descriptive string for a first data record having at least one word thereof with the second type of spelling; and
use the trained machine learning model to output a substitute descriptive string to replace the received descriptive string for the first data record, the substitute descriptive string having the first type of spelling of the at least one word instead of the second type of spelling.

### Claim 12 (depends on 11)

The health tracking system according to claim 11, wherein the data processor is further configured to:
generate a vector with the received descriptive string of the first data record using the machine learning model; and
identify the substitute descriptive string by applying the machine learning model to the generated vector.

### Claim 13 (depends on 11)

The health tracking system according to claim 11, wherein the data processor is further configured to:
filter the plurality of data records to identify the subset of the plurality of data records based on a frequency with which each data record in the plurality of data records has been selected by a plurality of users to be logged in a food diary.

### Claim 14 (depends on 11)

The health tracking system according to claim 11, wherein the data processor is further configured to:
filter the plurality of data records to identify the subset of the plurality of data records based on language settings for users who initially generated each data record in the plurality of data records.

### Claim 15 (depends on 11)

The health tracking system according to claim 11, wherein the data processor is further configured to:
generate the plurality of companion descriptive strings having at least one word thereof with the second type of spelling by introducing one or more spelling errors into the descriptive strings of the identified subset of the plurality of data records.

### Claim 16 (depends on 11)

The health tracking system according to claim 11, wherein the machine learning model includes an encoder having at least one long short term memory layer and a decoder having at least one long short term memory layer.

### Claim 17

A method of operating a health tracking system having a database configured to store a plurality of data records, each of the plurality of data records comprising at least a descriptive string and nutritional data regarding a respective consumable item, the method comprising:
receiving a first descriptive string having at least one word thereof spelled with a first spelling;
generating a data storage element with the first descriptive string using a machine learning model;
identifying a second descriptive string which has a second spelling of the at least one word by applying the machine learning model to the generated data storage element;
calculating a confidence factor regarding the second descriptive string using the machine learning model;
when it is determined that the confidence factor exceeds a predetermined threshold, searching the database to identify a first subset of data records in the plurality of data records having descriptive strings that are similar to the second descriptive string; and
when it is determined that the confidence factor is below the predetermined threshold, searching the database to identify a second subset of data records in the plurality of data records having descriptive strings that are similar to the first descriptive string.

### Claim 18 (depends on 17)

The method according to claim 17, wherein the act of generating the data storage element comprises:
generating a plurality of values corresponding to respective ones of a sequence of characters that comprises the at least one word of the first descriptive string; and
forming a vector using the generated plurality of values.

### Claim 19 (depends on 17)

The method according to claim 17, wherein the act of identifying the second descriptive string comprises:
determining, for each character of a sequence of characters that comprises the second descriptive string, a most probable correct character by applying the machine learning model to the generated data storage element; and
forming the second descriptive string using the determined most probable correct characters.

### Claim 20 (depends on 19)

The method according to claim 19, wherein the act of calculating the confidence factor comprises:
calculating the confidence factor based on probabilities of the determined most probable correct characters.
