Definition Extraction Using Dependency-Aware Machine Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing informational-extraction systems lack accuracy and flexibility in automatically extracting definitions from digital documents due to reliance on rule-based approaches or pre-specified features, failing to adapt to varying text structures and domains.

Innovation Solution

A definition extraction system utilizing a single machine-learning framework that classifies and labels word sequences based on global document structures and semantic consistencies, employing a multi-task-machine-learning model with graph convolutional neural networks to generate dependency paths between terms and definitions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If rule-based approaches or pre-specified features are used for definition extraction, then the system is simple to implement, but the accuracy and flexibility in extracting definitions from varying text structures and domains deteriorates

Engineering Contradiction:

Improveease of implementationVSAvoidextraction accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces rule-based mechanical extraction systems with a neural network-based machine learning system. The neural network automatically learns extraction patterns from training data, eliminating the need for manual rule creation while significantly improving extraction accuracy across different text structures and domains.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system changes from fixed pre-specified features to dynamic learned features. The neural network adapts its parameters through training on domain-specific data, allowing it to capture domain-specific terminology and patterns that pre-specified features cannot identify, thereby improving both accuracy and flexibility.

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If rule-based approaches are used for definition extraction, then the system has simple structure, but the adaptability to varying text structures and domains deteriorates

Engineering Contradiction:

Improvesystem structureVSAvoidadaptability to text structures

Core Design Contradiction:

Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent transforms the static rule-based system into a dynamic machine learning system. The neural network continuously adapts to different text structures and domains through training, automatically adjusting its extraction behavior based on learned patterns from domain-specific corpora, thereby achieving high versatility without requiring structural changes.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The neural network-based system achieves universal applicability across multiple domains and text structures through a single unified model. By training on diverse domain-specific data, the system becomes adaptable to various extraction tasks without requiring separate rule sets for each domain, thus improving versatility while maintaining reasonable structural complexity.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Productivity

If existing informational-extraction systems are used, then the processing speed is acceptable, but the accuracy in extracting definitions from complex sentences deteriorates

Engineering Contradiction:

Improveprocessing speedVSAvoiddefinition extraction accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent replaces traditional extraction mechanisms with neural network-based processing that can handle complex sentence structures more effectively. The neural network's ability to understand contextual relationships and semantic patterns enables accurate extraction from complex sentences while maintaining efficient processing speeds through optimized model architecture.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12536377B2Extracting definitions from documents utilizing definition-labeling-dependent machine learning background

Publication Date: 2026.01.27 ADOBE INC

AI summary

This disclosure describes methods, non-transitory computer readable storage media, and systems that extract a definition for a term from a source document by utilizing a single machine-learning framework to classify a word sequence from the source document as including a term definition and to label words from the word sequence. To illustrate, the disclosed system can receive a source document including a word sequence arranged in one or more sentences. The disclosed systems can utilize a machine-learning model to classify the word sequence as comprising a definition for a term and generate labels for the words from the word sequence corresponding to the term and the definition. Based on classifying the word sequence and the generated labels, the disclosed system can extract the definition for the term from the source document.