Word-Embedding Clustering for Believable Fake Document Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current cybersecurity systems are deficient in automatically selecting concepts for replacement in documents to generate believable yet diverse fake documents, often requiring expensive ontologies and failing to ensure diversity and randomness, making them susceptible to detection by adversaries.

Innovation Solution

A method utilizing word embeddings and clustering to identify potential replacements for concepts, incorporating joint optimization problems to ensure similarity and diversity in fake document generation, enhancing cybersecurity by generating credible yet distinct fake documents.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional deception technology is used to generate fake documents, then security detection capability is improved, but document diversity and randomness are insufficient making them susceptible to detection

Engineering Contradiction:

Improvesecurity detection capabilityVSAvoiddocument diversity

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system dynamically generates fake documents by replacing concepts with alternative terms from the same semantic cluster, creating varied document versions that adapt to different detection scenarios while maintaining believability

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes semantic parameters by substituting concepts with synonyms or related terms from word embedding clusters, transforming the original document into multiple versions with different lexical representations but preserved meaning

Inventive Principle:

Principle #35Parameter changes

2Reliability

If concept replacement is performed to generate fake documents, then document believability is improved, but detection precision by adversaries increases if patterns are predictable

Engineering Contradiction:

Improvedocument believabilityVSAvoidadversary detection precision

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The system creates multiple copies of the original document with concept replacements, generating fake documents that replicate the structure and style while substituting semantic content to maintain believability

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system combines original concepts with replacement concepts from word embedding clusters to create composite document versions that integrate familiar patterns with novel lexical choices

Inventive Principle:

Principle #40Composite materials

3Reliability

If manual examination of topics is performed to generate fake documents, then document realism is improved, but time consumption and difficulty to scale increase

Engineering Contradiction:

Improvedocument realismVSAvoidgeneration speed

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system replaces manual topic examination with automated word embedding and clustering algorithms, substituting human cognitive processes with computational methods that maintain realism while enabling scalable document generation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-service by automatically identifying concepts, generating replacements, and creating fake documents without manual intervention, enabling high-volume generation while preserving document quality

Inventive Principle:

Principle #25Self-service

4Reliability

If existing systems focus on numeric data changes, then structured field protection is improved, but technical document protection is insufficient

Engineering Contradiction:

Improvestructured field protectionVSAvoidtechnical document coverage

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system provides universal protection by extending concept replacement capabilities from numeric fields to unstructured text content, making the deception technology applicable to diverse document types including technical specifications and narratives

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12417506B2Generating fake documents using word embeddings to deter intellectual property theft

Publication Date: 2025.09.16 THE UNITED STATES OF AMERICA AS REPRESENTED BY THE SECRETARY OF THE NAVY

AI summary

A computer-implemented method, system and computer program product for generating fake documents. A corpus of domain specific documents is built and word embeddings for each word in such documents are identified as embedding vectors. Concepts in the corpus are then clustered together by clustering the embedding vectors. A feasible candidate replacement set is generated for each concept using the clustered concepts in the corpus. After such pre-processing steps are accomplished, concepts are extracted from a document. The concept importance values are computed for these extracted concepts, in which the extracted concepts are clustered into bins based on such measurements. A joint optimization problem is solved to identify both the concepts in the document to be replaced using the clustered concepts in the bins as well as the corresponding replacement concepts obtained from the clustered concepts in the corpus. Such replacements are made to generate a fake document.