# US 2026/0093748 A1 · Systems and methods for an ai framework for classifying labelled data

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

- Number: US 2026/0093748 A1
- Document ID: US20260093748A1
- Status: Published application
- Kind: A1
- Application number: 18/904,412
- Filed: 2024-10-02
- Published: 2026-04-02
- Earliest claimed: 2024-10-02
- Assignees: Verizon Patent and Licensing Inc.
- Applicants: Verizon Patent and Licensing Inc.
- Inventors: Rajkumar BONDUGULA; Deepa Thazhathu VEETIL
- Web page: https://patentgrep.com/patent/US20260093748A1

## CPC

- G06F 16/35: Clustering; Classification
- G06F 16/322: {Trees}
- G06F 16/3344: {using natural language analysis}
- G06F 16/383: using metadata automatically derived from the content
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Disclosed are systems and methods that provide a computerized framework for classifying data at scale, particularly when dealing with complex datasets that include ambiguous, incomplete and/or non-standard data. The disclosed framework provides a computerized solution that is scalable, accurate and capable of handling the complexities of the dataset while ensuring data treatment was done in line with legal and security requirements. The framework can operate to classify text-based data even when the data itself is incomplete, ambiguous, overlapping, and/or inconsistent, which can be particularly important for a wireless provider, where data is often scattered across different systems, and the information available for classification might be fragmented or contradictory. The framework provides functionality for classifying data automatically and efficiently, even as the volume of data grows exponentially.

## Claims

### Claim 1

A method comprising:
identifying a dataset;
analyzing the dataset, and determining, based on the analysis, a data profile and metadata;
executing a first computer model based on the data profile and metadata, and determining a classification of information within the dataset, the first computer model being trained to classify information based on statistical patterns and learned features;
executing a second computer model based on the data profile and metadata, and determining embeddings for the dataset, the embeddings capturing semantic relationships;
determining, based on the execution of the first computer model and the second computer model, a data classification, the data classification being based on the classification of the information within the dataset and the embeddings;
treating, based on the data classification, the dataset to generate a modified version of the dataset, the treatment comprising at least one of anonymizing or encrypting the dataset to enable sharing of the dataset across different environments in accordance with a policy; and
outputting, over a network, a device-consumable version of the modified dataset to a receiving entity, the outputting enabling usage of the modified dataset by a device of the receiving entity.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
annotating the dataset based on the classification of the information within the dataset and the embeddings, the annotations comprising a label corresponding to the information within the dataset.

### Claim 3 (depends on 2)

The method of claim 2, further comprising the modified dataset including the annotations.

### Claim 4

(canceled)

### Claim 5 (depends on 1)

The method of claim 1, further comprising the first computer model executing an artificial intelligence (AI) model.

### Claim 6 (depends on 5)

The method of claim 5, further comprising the AI model selected from a group consisting of: Naïve Bayes model, nearest neighbor model and decision tree model.

### Claim 7 (depends on 6)

The method of claim 6, further comprising the classification of information performed by the first computer model corresponding to a classification of columns within the dataset.

### Claim 8 (depends on 1)

The method of claim 1, further comprising execution of the second computer model comprising:
determining a Term Frequency Inverse Document Frequency (TF-IDF) representation;
executing a large language model (LLM); and
determining a classification of the dataset based on an output of the LLM and the TF-IDF representation.

### Claim 9

A system comprising:
a processor configured to:
identify a dataset;
analyze the dataset, and determine, based on the analysis, a data profile and metadata;
execute a first computer model based on the data profile and metadata, and determine a classification of information within the dataset, the first computer model being trained to classify information based on statistical patterns and learned features;
execute a second computer model based on the data profile and metadata, and determine embeddings for the dataset, the embeddings capturing semantic relationships;
determine, based on the execution of the first computer model and the second computer model, a data classification, the data classification being based on the classification of the information within the dataset and the embeddings;
treat, based on the data classification, the dataset to generate a modified version of the dataset, the treatment comprising at least one of anonymizing or encrypting the dataset to enable sharing of the dataset across different environments in accordance with a policy; and
output, over a network, a device-consumable version of the modified dataset to a receiving entity, the outputting enabling usage of the modified dataset by a device of the receiving entity.

### Claim 10 (depends on 9)

The system of claim 9, wherein the processor is further configured to:
annotate the dataset based on the classification of the information within the dataset and the embeddings, the annotations comprising a label corresponding to the information within the dataset.

### Claim 11 (depends on 9)

The system of claim 9, further comprising the modified dataset including the annotations.

### Claim 12

(canceled)

### Claim 13 (depends on 9)

The system of claim 9, further comprising the first computer model executing an artificial intelligence (AI) model.

### Claim 14 (depends on 13)

The system of claim 13, further comprising the AI model selected from a group consisting of: Naïve Bayes model, nearest neighbor model and decision tree model.

### Claim 15 (depends on 14)

The system of claim 14, further comprising the classification of information performed by the first computer model corresponding to a classification of columns within the dataset.

### Claim 16 (depends on 9)

The system of claim 9, wherein the processor is further configured to:
determine a Term Frequency Inverse Document Frequency (TF-IDF) representation;
execute a large language model (LLM); and
determine a classification of the dataset based on an output of the LLM and the TF-IDF representation.

### Claim 17

A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions, that when executed by a processor, perform a method comprising:
identifying a dataset;
analyzing the dataset, and determining, based on the analysis, a data profile and metadata;
executing a first computer model based on the data profile and metadata, and determining a classification of information within the dataset, the first computer model being trained to classify information based on statistical patterns and learned features;
executing a second computer model based on the data profile and metadata, and determining embeddings for the dataset, the embeddings capturing semantic relationships;
determining, based on the execution of the first computer model and the second computer model, a data classification, the data classification being based on the classification of the information within the dataset and the embeddings;
treating, based on the data classification, the dataset to generate a modified version of the dataset, the treatment comprising at least one of anonymizing or encrypting the dataset to enable sharing of the dataset across different environments in accordance with a policy; and
outputting, over a network, a device-consumable version of the modified dataset to a receiving entity, the outputting enabling usage of the modified dataset by a device of the receiving entity.

### Claim 18 (depends on 17)

The non-transitory computer-readable storage medium of claim 17, further comprising:
annotating the dataset based on the classification of the information within the dataset and the embeddings, the annotations comprising a label corresponding to the information within the dataset.

### Claim 19

(canceled)

### Claim 20 (depends on 17)

The non-transitory computer-readable storage medium of claim 17, further comprising:
determining a Term Frequency Inverse Document Frequency (TF-IDF) representation;
executing a large language model (LLM); and
determining a classification of the dataset based on an output of the LLM and the TF-IDF representation, wherein the data classification is based on the determined classification of the dataset based on the output of the LLM, TF-IDF and the first model.

### Claim 21 (depends on 1)

The method of claim 1, wherein the treatment includes at least one of one-way hashing, two-way hashing, or masking.

### Claim 22 (depends on 9)

The system of claim 9, wherein the treatment includes at least one of one-way hashing, two-way hashing, masking, or encryption.

### Claim 23 (depends on 17)

The non-transitory computer-readable storage medium of claim 17, wherein the treatment includes at least one of one-way hashing, two-way hashing, masking, or encryption.
