# US 12,001,789 B2 · Textual data analytics

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

- Number: US 12,001,789 B2
- Document ID: US12001789B2
- Status: Granted
- Kind: B2
- Application number: 17/446,220
- Filed: 2021-08-27
- Published: 2023-03-02
- Granted: 2024-06-04
- Earliest claimed: 2021-08-27
- Assignees: S&P Global Inc.
- Applicants: S&P Global Inc.
- Inventors: Fanxi Zhao; David Pope; Daniel Sandberg; Temilade Oyeniyi; Ashish Sumsher Rana; Eric Oak
- Web page: https://patentgrep.com/patent/US12001789B2

## CPC

- G06F 40/216: using statistical methods
- G06N 5/045: Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
- G06F 40/279: Recognition of textual entities
- G06N 5/04: Inference or reasoning models
- G06Q 30/0202: Market predictions or forecasting for commercial activities
- G06Q 40/06: Asset management; Financial planning or analysis
- G06F 40/205: Parsing
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

A method of textual data analysis is provided. The method comprises parsing text data extracted from transcripts related to a number of companies. Intermediate metrics are created comprising numerical representations of the parsed text data and derivations. The intermediate metrics are then combined into different combinations comprising headline analytics. A machine learning model tests each headline analytic for standalone predictive efficacy. Headline analytics with standalone predictive efficacies above a first threshold are selected and then tested for additive predictive efficacy to determine if the selected headline analytic incrementally increases the predictive efficacy of a preexisting economic analytic above a second threshold. Headline analytics with additive predictive efficacy above the second threshold are applied to a second number of transcripts in combination with the preexisting economic analytic to predict financial performance of companies that are the subjects of the second number of transcripts.

## Claims

### Claim 1

A computer-implemented method of textual data analysis, the method comprising:
using a number of processors to perform the steps of:
parsing, according to natural language processing, text data extracted from a first number of transcripts related to a number of companies;
creating a number of intermediate metrics comprising numerical representations of the parsed text data and derivations of the parsed text data;
combining the intermediate metrics into a number of different combinations, wherein each combination comprises a headline analytic;
testing, according to a machine learning model, each headline analytic for standalone predictive efficacy based on historical data;
selecting headline analytics with standalone predictive efficacies above a first threshold;
testing, according to the machine learning model, each selected headline analytic for additive predictive efficacy to determine if the selected headline analytic incrementally increases the predictive efficacy of a preexisting economic analytic above a second threshold based on historical data; and
applying, to a second number of transcripts, the selected headline analytics with an additive predictive efficacy above the second threshold, in combination with the preexisting economic analytic, to predict financial performance of companies that are the subjects of the second number of transcripts.

### Claim 2 (depends on 1)

The method of claim 1, wherein testing the selected headline analytics for additive predictive efficacy further comprises checking the additive predictive efficacy of each selected headline analytic for robustness.

### Claim 3 (depends on 2)

The method of claim 2, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific industry sector.

### Claim 4 (depends on 2)

The method of claim 2, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a particular calendar year or quarter.

### Claim 5 (depends on 2)

The method of claim 2, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific construction wherein the predictive efficacy disappears with one or more modifications to the construction.

### Claim 6 (depends on 1)

The method of claim 1, wherein the extracted text data comprise transcriptions of speech spoken by company executives and financial analysts.

### Claim 7 (depends on 1)

The method of claim 1, wherein parsing the extracted text data comprises parsing the text into sentences that serve as base units of analysis, wherein the sequential order of sentences is preserved from the transcripts.

### Claim 8 (depends on 1)

The method of claim 1, further comprising cleaning the extracted text data by performing at least one of:
expanding contractions;
removing punctuations;
removing stop words;
filtering on a master word list;
making all letters lower case; or
tokenizing sentences.

### Claim 9 (depends on 1)

The method of claim 1, wherein the transcripts comprise earnings call transcripts.

### Claim 10 (depends on 1)

The method of claim 1, further comprising removing data biases from the extracted text data before creating the intermediate metrics and headline analytics.

### Claim 11 (depends on 10)

The method of claim 10, wherein data biases might comprise at least one of:
look-ahead bias; or
survivorship bias.

### Claim 12 (depends on 1)

The method of claim 1, wherein each headline analytic might comprise one of:
market moving topics and neighboring descriptors;
weighted average sentiment score; or
gauging call level transparency.

### Claim 13

A system for textual data analysis, the system comprising:
a storage device configured to store program instructions; and
one or more processors operably connected to the storage device and configured to execute the program instructions to cause the system to:
parsing, according to natural language processing, text data extracted from a first number of transcripts related to a number of companies;
create a number of intermediate metrics comprising numerical representations of the parsed text data and derivations of the parsed text data;
combine the intermediate metrics into a number of different combinations, wherein each combination comprises a headline analytic;
test, according to a machine learning model, each headline analytic for standalone predictive efficacy based on historical data;
select headline analytics with standalone predictive efficacies above a first threshold;
test, according to the machine learning model, each selected headline analytic for additive predictive efficacy to determine if the selected headline analytic incrementally increases the predictive efficacy of a preexisting economic analytic above a second threshold based on historical data; and
apply, to a second number of transcripts, the selected headline analytics with an additive predictive efficacy above the second threshold, in combination with the preexisting economic analytic, to predict financial performance of companies that are the subjects of the second number of transcripts.

### Claim 14 (depends on 13)

The system of claim 13, wherein testing the selected headline analytics for additive predictive efficacy further comprises checking the additive predictive efficacy of each selected headline analytic for robustness.

### Claim 15 (depends on 14)

The system of claim 14, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific industry sector.

### Claim 16 (depends on 14)

The system of claim 14, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a particular calendar year or quarter.

### Claim 17 (depends on 14)

The system of claim 14, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific construction wherein the predictive efficacy disappears with one or more modifications to the construction.

### Claim 18 (depends on 13)

The system of claim 13, wherein the extracted text data comprise transcriptions of speech spoken by company executives and financial analysts.

### Claim 19 (depends on 13)

The system of claim 13, wherein parsing the extracted text data comprises parsing the text into sentences that serve as base units of analysis, wherein the sequential order of sentences is preserved from the transcripts.

### Claim 20 (depends on 13)

The system of claim 13, wherein the processors further execute instructions for cleaning the extracted text data by performing at least one of:
expanding contractions;
removing punctuations;
removing stop words;
filtering on a master word list;
making all letters lower case; or
tokenizing sentences.

### Claim 21 (depends on 13)

The system of claim 13, wherein the transcripts comprise earnings call transcripts.

### Claim 22 (depends on 13)

The system of claim 13, wherein the processors further execute instructions for removing data biases from the extracted text data before creating the intermediate metrics and headline analytics.

### Claim 23 (depends on 13)

The system of claim 13, wherein data biases might comprise at least one of:
look-ahead bias; or
survivorship bias.

### Claim 24 (depends on 13)

The system of claim 13, wherein each headline analytic might comprise one of:
market moving topics and neighboring descriptors;
weighted average sentiment score; or
gauging call level transparency.

### Claim 25

A computer program product for textual data analysis, the computer program product comprising:
a computer-readable storage medium having program instructions embodied thereon to perform the steps of:
parsing, according to natural language processing, text data extracted from a first number of transcripts related to a number of companies;
creating a number of intermediate metrics comprising numerical representations of the parsed text data and derivations of the parsed text data;
combining the intermediate metrics into a number of different combinations, wherein each combination comprises a headline analytic;
testing, according to a machine learning model, each headline analytic for standalone predictive efficacy based on historical data;
selecting headline analytics with standalone predictive efficacies above a first threshold;
testing, according to the machine learning model, each selected headline analytic for additive predictive efficacy to determine if the selected headline analytic incrementally increases the predictive efficacy of a preexisting economic analytic above a second threshold based on historical data; and
applying, to a second number of transcripts, the selected headline analytics with an additive predictive efficacy above the second threshold, in combination with the preexisting economic analytic, to predict financial performance of companies that are the subjects of the second number of transcripts.

### Claim 26 (depends on 25)

The computer program product of claim 25, wherein testing the selected headline analytics for additive predictive efficacy further comprises checking the additive predictive efficacy of each selected headline analytic for robustness.

### Claim 27 (depends on 26)

The computer program product of claim 26, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific industry sector.

### Claim 28 (depends on 26)

The computer program product of claim 26, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a particular calendar year or quarter.

### Claim 29 (depends on 26)

The computer program product of claim 26, wherein checking the additive predictive efficacy of each selected headline analytic for robustness comprises determining if the additive predictive efficacy is limited to a specific construction wherein the predictive efficacy disappears with one or more modifications to the construction.

### Claim 30 (depends on 25)

The computer program product of claim 25, wherein the extracted text data comprise transcriptions of speech spoken by company executives and financial analysts.

### Claim 31 (depends on 25)

The computer program product of claim 25, wherein parsing the extracted text data comprises parsing the text into sentences that serve as base units of analysis, wherein the sequential order of sentences is preserved from the transcripts.

### Claim 32 (depends on 25)

The computer program product of claim 25, further comprising instructions for cleaning the extracted text data by performing at least one of:
expanding contractions;
removing punctuations;
removing stop words;
filtering on a master word list;
making all letters lower case; or
tokenizing sentences.

### Claim 33 (depends on 25)

The computer program product of claim 25, wherein the transcripts comprise earnings call transcripts.

### Claim 34 (depends on 25)

The computer program product of claim 25, further comprising instructions for removing data biases from the extracted text data before creating the intermediate metrics and headline analytics.

### Claim 35 (depends on 25)

The computer program product of claim 25, wherein data biases might comprise at least one of:
look-ahead bias; or
survivorship bias.

### Claim 36 (depends on 25)

The computer program product of claim 25, wherein each headline analytic might comprise one of:
market moving topics and neighboring descriptors;
weighted average sentiment score; or
gauging call level transparency.