# US 2021/0342640 A1 · Automated machine-learning dataset preparation

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

- Number: US 2021/0342640 A1
- Document ID: US20210342640A1
- Status: Published application
- Kind: A1
- Application number: 16/865,621
- Filed: 2020-05-04
- Published: 2021-11-04
- Earliest claimed: 2020-05-04
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Willie Robert Patten, JR.; Eugene Irving Kelton; Arvin Bhatnagar; Jason Howard Cornpropst; Jacob McPherson
- Web page: https://patentgrep.com/patent/US20210342640A1

## CPC

- G06F 40/279: Recognition of textual entities
- G06F 16/9024: {Graphs; Linked lists (G06F16/9027 takes precedence)}
- G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade}
- G06F 18/2193: {based on specific statistical tests}
- G06F 40/205: Parsing
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A method of preparing a dataset may comprise calculating a pattern relevance for a first field in the dataset and modifying the first field based on the pattern relevance. The method may further comprise detecting a contextual cue in the first field. The method may further comprise retrieving contextual information for a value in the first field and adding that contextual information to the database. Finally, the method may further comprise identifying a numerical scheme for the first field and parsing the first field into a number according to that numerical scheme.

## Claims

### Claim 1

A method of preparing a dataset for ingestion by a machine-learning model, the method comprising:
calculating a pattern relevance for a first field in the dataset;
modifying the first field based on the pattern relevance;
detecting a contextual cue in the first field;
retrieving contextual information for a value in the first field;
adding the contextual information to the dataset;
identifying a numerical scheme for the first field; and
parsing the first field into a number according to the numerical scheme.

### Claim 2 (depends on 1)

The method of claim 1, wherein calculating the pattern relevance comprises:
calculating a variance for the first field; and
comparing the variance to a variance threshold, wherein the modifying is based on the comparison.

### Claim 3 (depends on 2)

The method of claim 2, wherein the modifying comprises removing the first field from the data set.

### Claim 4 (depends on 1)

The method of claim 1, wherein calculating the pattern relevance comprises determining that the first field is linked with a second field in the dataset, wherein the modifying includes merging the linked fields.

### Claim 5 (depends on 1)

The method of claim 1, wherein calculating the pattern relevance comprises determining that the first field is linked with a second field in the dataset, wherein the modifying includes adding a third field to the dataset that expresses the linked relationship.

### Claim 6 (depends on 1)

The method of claim 1, wherein calculating the pattern relevance comprises performing natural-language processing on communications associated with the dataset.

### Claim 7 (depends on 1)

The method of claim 1, further comprising determining a number of potential variants for the first field, wherein the numerical scheme is based on the number of potential variants.

### Claim 8 (depends on 1)

The method of claim 1, wherein the contextual cue is a likelihood that contextual information related to the first field may be available and may be of use in associating patterns in the dataset with a target phenomenon.

### Claim 9 (depends on 1)

The method of claim 1, wherein adding the contextual information to the dataset comprises adding a fourth field to the dataset that contains the contextual information.

### Claim 10

A system comprising:
a processor; and
a memory in communication with the processor, the memory containing program instructions that, when executed by the processor, are configured to cause the processor to perform a method of preparing a dataset for ingestion by a machine-learning model, the method comprising:
calculating a pattern relevance for a first field in the dataset;
modifying the first field based on the pattern relevance;
detecting a contextual cue in the first field;
retrieving contextual information for a value in the first field;
adding the contextual information to the dataset;
identifying a numerical scheme for the first field; and
parsing the first field into a number according to the numerical scheme.

### Claim 11 (depends on 10)

The system of claim 10, wherein calculating the pattern relevance comprises:
calculating a variance for the first field; and
comparing the variance to a variance threshold, wherein the modifying is based on the comparison.

### Claim 12 (depends on 10)

The system of claim 10, wherein calculating the pattern relevance comprises determining that the first field is linked with a second field in the dataset, wherein the modifying includes adding a third field to the dataset that expresses the linked relationship.

### Claim 13 (depends on 10)

The system of claim 10, wherein the method further comprises determining a number of potential variants for the first field, and wherein the numerical scheme is based on the number of potential variants.

### Claim 14 (depends on 10)

The system of claim 10, wherein the contextual cue is a likelihood that contextual information related to the first field may be available and may be of use in associating patterns in the dataset with a target phenomenon.

### Claim 15 (depends on 10)

The system of claim 10, wherein adding the contextual information to the dataset comprises adding a fourth field to the dataset that contains the contextual information.

### Claim 16

A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to:
calculate a pattern relevance for a first field in a dataset;
modify the first field based on the pattern relevance;
detect a contextual cue in the first field;
retrieve contextual information for a value in the first field;
add the contextual information to the dataset;
identify a numerical scheme for the first field; and
parse the first field into a number according to the numerical scheme.

### Claim 17 (depends on 16)

The computer program product of claim 16, wherein calculating the pattern relevance comprises:
calculating a variance for the first field; and
comparing the variance to a variance threshold, wherein the modifying is based on the comparison.

### Claim 18 (depends on 16)

The computer program product of claim 16, wherein calculating the pattern relevance comprises determining that the first field is linked with a second field in the dataset, wherein the modifying includes adding a third field to the dataset that expresses the linked relationship.

### Claim 19 (depends on 16)

The computer program product of claim 16, wherein the program instructions further cause the computer to determine a number of potential variants for the first field, and wherein the numerical scheme is based on the number of potential variants.

### Claim 20 (depends on 16)

The computer program product of claim 16, wherein the contextual cue is a likelihood that contextual information related to the first field may be available and may be of use in associating patterns in the dataset with a target phenomenon.
