Variable Lexical Classes for Word Sense Disambiguation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional techniques for learning regular structures over natural languages rely on ontological categories, which hinder accuracy, and fixed lexical features, limiting the prediction of ambiguous terms, especially when using 'bags of words'.

Innovation Solution

The system employs variable lexical classes to identify relevant context for word sense disambiguation, representing lexical information through clustering specific contexts, allowing an indefinite number of lexical features and creating regular rules based on both lexical and syntactical features from a corpus of sentences.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If ontological categories are used to extract regular expressions from corpora, then the method can be applied to semantic processing, but the accuracy is hindered

Engineering Contradiction:

Improveapplicability to semantic processingVSAvoidaccuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent changes the parameter from fixed ontological categories to variable lexical classes. This allows the system to adapt to different lexical contexts dynamically, improving accuracy while maintaining applicability to semantic processing tasks.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces dynamic lexical class assignment where words are assigned to lexical classes based on their contextual usage patterns. This dynamic approach allows the system to capture semantic nuances that static ontological categories miss, thereby improving accuracy.

Inventive Principle:

Principle #15Dynamics

2Ease of manufacture

If bags of words with fixed number of lexical features are used, then the method is simple to implement, but the prediction of ambiguous terms is limited

Engineering Contradiction:

Improvesimplicity of implementationVSAvoidprediction accuracy of ambiguous terms

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent transitions from fixed-number lexical features to variable lexical features where the number and selection of features depend on the contextual requirements. This dynamic feature selection enables accurate prediction of ambiguous terms while maintaining implementation simplicity through automated feature generation.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter from fixed lexical feature count to variable lexical feature count based on context. This allows the system to use more features when needed for ambiguous terms and fewer features when context is clear, improving prediction accuracy without significantly complicating implementation.

Inventive Principle:

Principle #35Parameter changes

3Stability of the object's composition

If fixed lexical features are used, then the system is stable and consistent, but the accuracy increases significantly when using variable lexical classes

Engineering Contradiction:

Improvesystem stabilityVSAvoidaccuracy

Core Design Contradiction:

Stability of the object's compositionVSMeasurement precision

Solution Approach 1:

The patent introduces variable lexical classes that adapt to different contexts while maintaining a stable underlying framework. The system generates lexical classes dynamically based on corpus analysis, ensuring both stability in the core system architecture and flexibility in handling diverse linguistic patterns for improved accuracy.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10169333B2System, method, and recording medium for regular rule learning

Publication Date: 2019.01.01 AIRBNB INC

AI summary

A regular rule learning system, including an analyzing circuit configured to analyze a corpus of sentences to find semantic relationships between sentence constituents that are responsible for specific senses of words in that sentence by describing the semantic relationships and grammatical relations that are actuated in the sentence.