# US 2026/0086992 A1 · Deep-Learning Solution to Building First-Party Identity Graphs from Third-Party Data

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

- Number: US 2026/0086992 A1
- Document ID: US20260086992A1
- Status: Published application
- Kind: A1
- Application number: 19/397,184
- Filed: 2025-11-21
- Published: 2026-03-26
- Earliest claimed: 2024-09-24
- Assignees: LiveRamp, Inc.
- Applicants: LiveRamp, Inc.
- Inventors: Venkatesh Ramanathan; Jaipal Nenavath
- Web page: https://patentgrep.com/patent/US20260086992A1

## CPC

- G06F 16/2237: {Vectors, bitmaps or matrices}
- H04L 9/0894: {Escrow, recovery or storing of secret information, e.g. secret key escrow or cryptographic key storage}
- G06F 16/215: Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
- G06F 16/22: Indexing; Data structures therefor; Storage structures
- G06F 16/2468: {Fuzzy queries}
- G06F 16/248: Presentation of query results
- G06F 16/33: Querying
- G06F 16/56: having vectorial format
- G06F 16/901: Indexing; Data structures therefor; Storage structures (for retrieval from the web G06F16/951)
- G06F 21/6245: {Protecting personal data, e.g. for financial or medical purposes}
- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/295: Named entity recognition
- G06F 40/30: Semantic analysis
- G06N 3/02: Neural networks
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/048: Activation functions
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 3/096: Transfer learning
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06N 5/048: Fuzzy inferencing
- G06N 20/00: Machine learning
- G06N 20/20: Ensemble learning
- G06F 16/25: Integrating or interfacing systems involving database management systems
- G06F 40/10: Text processing (natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40)
- G06N 5/02: Knowledge representation; Symbolic representation
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system and method for building first-party identity graphs using third-party data employs deep learning to enable privacy-preserving identity resolution. The system includes a deep learning model trained with transformer architecture and contrastive learning on a comprehensive third-party identity graph containing personally identifiable information (PII). Custom tokenizers process PII data by leveraging hierarchical structures and domain-specific characteristics to handle variations in spellings, abbreviations, and data formats. An identity matcher generates vector embeddings from first-party PII inputs, which are stored in a vector database for efficient similarity searches. A distributed similarity computation component compares first-party embeddings with third-party data embeddings, outputting similarity scores that enable fuzzy matching capabilities. A graph builder constructs accurate first-party identity graphs based on similarity thresholds while incorporating relevant third-party data.

## Claims

### Claim 1

A system for building first-party identity graphs using third-party data, comprising:
a deep learning model trained on a third-party identity graph;
an identity matcher configured to:
receive first-party personally identifiable information (PII) inputs; and
generate embeddings in the form of vectors of floating point values based on the first-party PII inputs using the deep learning model;
a vector database configured to store, index, and retrieve vector data corresponding to the first-party PII embeddings;
a distributed similarity computation component configured to:
take as input the first-party PII embeddings;
compare the first-party PII embeddings with third-party data embeddings; and
output similarity scores between first-party and third-party data points; and
a graph builder configured to:
construct a first-party identity graph based on the similarity scores and a predefined threshold; and
incorporate relevant third-party data into the first-party identity graph.

### Claim 2 (depends on 1)

The system of claim 1, wherein the deep learning model is trained using contrastive learning on the third-party identity graph to improve differentiation between similar but distinct identities.

### Claim 3 (depends on 1)

The system of claim 1, wherein the identity matcher is configured to generate embeddings that preserve similarities between different representations of the same identity across first-party and third-party data.

### Claim 4 (depends on 1)

The system of claim 1, further comprising a custom tokenizer configured to process first-party PII data, including names with variations in format and spelling, email addresses containing name information, addresses with abbreviations and variations, and telephone numbers with hierarchical information.

### Claim 5 (depends on 1)

The system of claim 1, wherein the distributed similarity computation component is configured to efficiently process large volumes of first-party and third-party identity data.

### Claim 6 (depends on 1)

The system of claim 1, wherein the graph builder is configured to apply context-specific criteria when incorporating third-party data into the first-party identity graph.

### Claim 7 (depends on 1)

The system of claim 1, further comprising a flexible schema handler configured to process input data without adhering to a strict schema, allowing for variations in data structure and content across first-party and third-party sources.

### Claim 8 (depends on 1)

The system of claim 1, wherein the system is configured to build the first-party identity graph without requiring movement of raw first-party PII data outside a secure environment.

### Claim 9 (depends on 1)

The system of claim 1, further comprising a distributed cluster ID generation component configured to generate unique identifiers for entities that appear in both first-party and third-party data.

### Claim 10

A method for building first-party identity graphs using third-party data, comprising:
receiving first-party personally identifiable information (PII) inputs;
generating embeddings in the form of vectors of floating point values based on the first-party PII inputs using a deep learning model trained on a third-party identity graph;
storing, indexing, and retrieving vector data corresponding to the first-party PII embeddings in a vector database;
performing distributed similarity computation by:
taking as input the first-party PII embeddings;
comparing the first-party PII embeddings with third-party data embeddings; and
outputting similarity scores between first-party and third-party data points; constructing a first-party identity graph based on the similarity scores; and incorporating relevant third-party data into the first-party identity graph.

### Claim 11 (depends on 10)

The method of claim 10, further comprising training the deep learning model using contrastive learning on the third-party identity graph to improve differentiation between similar but distinct identities.

### Claim 12 (depends on 10)

The method of claim 10, wherein generating embeddings preserves similarities between different representations of the same identity across first-party and third-party data.

### Claim 13 (depends on 10)

The method of claim 10, further comprising tokenizing the first-party PII inputs using a custom tokenizer configured to process names with variations in format and spelling, email addresses containing name information, addresses with abbreviations and variations, and telephone numbers with hierarchical information.

### Claim 14 (depends on 10)

The method of claim 10, wherein performing distributed similarity computation includes efficiently processing large volumes of first-party and third-party identity data.

### Claim 15 (depends on 10)

The method of claim 10, wherein incorporating relevant third-party data into the first-party identity graph includes applying context-specific criteria.

### Claim 16 (depends on 10)

The method of claim 10, further comprising processing input data without adhering to a strict schema, thus allowing for variations in data structure and content across first-party and third-party sources.

### Claim 17 (depends on 10)

The method of claim 10, further comprising building the first-party identity graph without requiring movement of raw first-party PII data outside a secure environment.

### Claim 18 (depends on 10)

The method of claim 10, further comprising generating unique identifiers for entities that appear in both first-party and third-party data using a distributed cluster ID generation component.

### Claim 19 (depends on 10)

The method of claim 10, further comprising updating the first-party identity graph incrementally as new first-party data becomes available, using the deep learning model to generate embeddings for the new data and incorporating it into the existing graph structure.

### Claim 20 (depends on 10)

The method of claim 10, further comprising using the first-party identity graph to enhance first-party data with relevant third-party information while maintaining privacy and security of the underlying PII data.