# US 2023/0066971 A1 · Corporate Hierarchy Tagging

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

- Number: US 2023/0066971 A1
- Document ID: US20230066971A1
- Status: Published application
- Kind: A1
- Application number: 17/446,509
- Filed: 2021-08-31
- Published: 2023-03-02
- Earliest claimed: 2021-08-31
- Assignees: S&P Global Inc.
- Applicants: S&P Global Inc.
- Inventors: Brian Tsay; Jason Prentice; James Ryan Psota
- Web page: https://patentgrep.com/patent/US20230066971A1

## CPC

- G06Q 10/087: Inventory or stock management, e.g. order filling, procurement or balancing against orders
- G06F 16/90335: {Query processing}
- G06F 16/906: Clustering; Classification
- G06F 18/213: Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 20/20: Ensemble learning
- G06N 3/084: Backpropagation, e.g. using gradient descent
- 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
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A method for identifying a parent company in a corporate hierarchy is provided. A neural network embeds a number of company names into respective numeric vectors and then indexes the vectors into an approximate nearest neighbors (ANN) index according to euclidean distances between them. In response to a query regarding a parent company of a specified query company, a top N number of nearest neighbor companies to the query company is extracted from the ANN index. A machine learning voting model determines which, if any, of the extracted nearest neighbor companies has a parent company that best corresponds to the query company. If an extracted nearest neighbor company has a parent company that best corresponds to the query company, the parent company is displayed to a user through a user interface.

## Claims

### Claim 1

A computer-implemented method for identifying a parent company in a corporate hierarchy, the method comprising:
using a number of processors to perform the steps of:
converting, by a neural network, a number of company names into respective numeric vectors;
indexing the numeric vectors into an approximate nearest neighbors (ANN) index according to euclidean distances between the vectors;
receiving a query regarding a specified query company within the number of company names, wherein the query is for a parent company of the query company;
in response to the query, extracting from the ANN index a top N number of nearest neighbor companies to the query company;
determining, by a machine learning voting model, which, if any, of the extracted nearest neighbor companies has a parent company that best corresponds to the query company; and
if an extracted nearest neighbor company has a parent company that best corresponds to the query company, displaying the parent company to a user through a user interface.

### Claim 2 (depends on 1)

The method of claim 1, wherein the neural network is trained with a triplet-based loss.

### Claim 3 (depends on 2)

The method of claim 2, wherein distance between an anchor and a positive example is made, by training, to be shorter than the distance between the anchor and a negative example.

### Claim 4 (depends on 1)

The method of claim 1, wherein the neural network is trained by randomly transforming company names.

### Claim 5 (depends on 4)

The method of claim 4, wherein transforming company names comprises at least one of:
adding common shipping terms to the beginning of a company name;
adding common shipping terms to the end of a company name;
changing the order of words in a company name;
deleting random letters from a company name; or
adding random spaces in a company name.

### Claim 6 (depends on 1)

The method of claim 1, wherein the voting model converts the extracted nearest neighbor companies into a set of features comprising at least one of:
raw distance;
binary indication whether the closest of the extracted neighbors belongs to a modal corporate hierarchy;
a proportion of the extracted nearest neighbors that belong to the modal corporate hierarchy;
average distance between the extracted nearest neighbors that belong to the modal corporate hierarchy; or
the distance between the query company and the closest of the extracted neighbors that belongs to the modal corporate hierarchy.

### Claim 7 (depends on 1)

The method of claim 1, wherein the voting model outputs one of the following predictions:
the parent company of the query company is the parent company of the nearest neighbor company;
the parent company of the query company is a modal parent company in a returned set of neighbor companies; or
no corporate hierarchy can be found to which the query company belongs among the nearest neighbor companies.

### Claim 8 (depends on 7)

The method of claim 7, wherein, if there is no mode among the neighbor companies, the voting model assumes a parent company of the farthest away neighbor company is the modal parent.

### Claim 9

A system for identifying a parent company in a corporate hierarchy, 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:
convert, by a neural network, a number of company names into respective numeric vectors;
index the numeric vectors into an approximate nearest neighbors (ANN) index according to euclidean distances between the vectors;
receive a query regarding a specified query company within the number of company names, wherein the query is for a parent company of the query company;
in response to the query, extract from the ANN index a top N number of nearest neighbor companies to the query company;
determine, by a machine learning voting model, which, if any, of the extracted nearest neighbor companies has a parent company that best corresponds to the query company; and
if an extracted nearest neighbor company has a parent company that best corresponds to the query company, display the parent company to a user through a user interface.

### Claim 10 (depends on 9)

The system of claim 9, wherein the neural network is trained with a triplet-based loss.

### Claim 11 (depends on 9)

The system of claim 9, wherein distance between an anchor and a positive example is made, by training, to be shorter than the distance between the anchor and a negative example.

### Claim 12 (depends on 9)

The system of claim 9, wherein the neural network is trained by randomly transforming company names.

### Claim 13 (depends on 12)

The system of claim 12, wherein transforming company names comprises at least one of:
adding common shipping terms to the beginning of a company name;
adding common shipping terms to the end of a company name;
changing the order of words in a company name;
deleting random letters from a company name; or
adding random spaces in a company name.

### Claim 14 (depends on 9)

The system of claim 9, wherein the voting model converts the extracted nearest neighbor companies into a set of features comprising at least one of:
raw distance;
binary indication whether the closest of the extracted neighbors belongs to a modal corporate hierarchy;
a proportion of the extracted nearest neighbors that belong to the modal corporate hierarchy;
average distance between the extracted nearest neighbors that belong to the modal corporate hierarchy; or
the distance between the query company and the closest of the extracted neighbors that belongs to the modal corporate hierarchy.

### Claim 15 (depends on 9)

The system of claim 9, wherein the voting model outputs one of the following predictions:
the parent company of the query company is the parent company of the nearest neighbor company;
the parent company of the query company is a modal parent company in a returned set of neighbor companies; or
no corporate hierarchy can be found to which the query company belongs among the nearest neighbor companies.

### Claim 16 (depends on 9)

The system of claim 9, wherein, if there is no mode among the neighbor companies, the voting model assumes a parent company of the farthest away neighbor company is the modal parent.

### Claim 17

A computer program product for identifying a parent company in a corporate hierarchy, the computer program product comprising:
a computer-readable storage medium having program instructions embodied thereon to perform the steps of:
converting, by a neural network, a number of company names into respective numeric vectors;
indexing the numeric vectors into an approximate nearest neighbors (ANN) index according to euclidean distances between the vectors;
receiving a query regarding a specified query company within the number of company names, wherein the query is for a parent company of the query company;
in response to the query, extracting from the ANN index a top N number of nearest neighbor companies to the query company;
determining, by a machine learning voting model, which, if any, of the extracted nearest neighbor companies has a parent company that best corresponds to the query company; and
if an extracted nearest neighbor company has a parent company that best corresponds to the query company, displaying the parent company to a user through a user interface.

### Claim 18 (depends on 17)

The computer program product of claim 17, wherein the neural network is trained with a triplet-based loss.

### Claim 19 (depends on 17)

The computer program product of claim 17, wherein distance between an anchor and a positive example is made, by training, to be shorter than the distance between the anchor and a negative example.

### Claim 20 (depends on 17)

The computer program product of claim 17, wherein the neural network is trained by randomly transforming company names.

### Claim 21 (depends on 20)

The computer program product of claim 20, wherein transforming company names comprises at least one of:
adding common shipping terms to the beginning of a company name;
adding common shipping terms to the end of a company name;
changing the order of words in a company name;
deleting random letters from a company name; or
adding random spaces in a company name.

### Claim 22 (depends on 17)

The computer program product of claim 17, wherein the voting model converts the extracted nearest neighbor companies into a set of features comprising at least one of:
raw distance;
binary indication whether the closest of the extracted neighbors belongs to a modal corporate hierarchy;
a proportion of the extracted nearest neighbors that belong to the modal corporate hierarchy;
average distance between the extracted nearest neighbors that belong to the modal corporate hierarchy; or
the distance between the query company and the closest of the extracted neighbors that belongs to the modal corporate hierarchy.

### Claim 23 (depends on 17)

The computer program product of claim 17, wherein the voting model outputs one of the following predictions:
the parent company of the query company is the parent company of the nearest neighbor company;
the parent company of the query company is a modal parent company in a returned set of neighbor companies; or
no corporate hierarchy can be found to which the query company belongs among the nearest neighbor companies.

### Claim 24 (depends on 17)

The computer program product of claim 17, wherein, if there is no mode among the neighbor companies, the voting model assumes a parent company of the farthest away neighbor company is the modal parent.
