Neural Network Translation Verification via Word Association

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Translation errors often occur when translating isolated words, such as UI menu items, due to the lack of proper context, requiring significant manual effort for identification and correction.

Innovation Solution

A neural network is trained to determine association degrees among groups of words in source and target languages, allowing for the verification and correction of translations by identifying wrong translations and suggesting correct alternatives.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual verification of translations is performed, then translation accuracy can be ensured, but significant manual effort and time are required

Engineering Contradiction:

Improvetranslation accuracyVSAvoidmanual effort time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The translation system performs self-verification by using the trained neural network to automatically check translation quality and identify errors without requiring manual human review, thereby maintaining accuracy while eliminating manual effort

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual verification process with an automated neural network-based verification system that uses association degree calculations to detect translation errors automatically

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If translation verification is performed without context, then processing speed is maintained, but translation errors occur frequently

Engineering Contradiction:

Improveprocessing speedVSAvoidtranslation accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent introduces association degree as an intermediary metric that mediates between source words and target translations, enabling the system to verify translations automatically while maintaining both speed and accuracy through contextual relationship analysis

Inventive Principle:

Principle #24Intermediary (Mediator)

3Extent of automation

If automated translation verification is implemented, then manual effort is reduced, but complex neural network training and computation are required

Engineering Contradiction:

Improveautomatic verification capabilityVSAvoidneural network complexity

Core Design Contradiction:

Extent of automationVSDevice complexity

Solution Approach 1:

The patent segments the translation verification process into distinct phases: data preparation, neural network training with specific loss functions, and inference with association degree calculation. This segmentation makes the complex system more manageable and implementable

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS11966711B2Translation verification and correction

Publication Date: 2024.04.23 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Embodiments of the present disclosure relate to a solution for translation verification and correction. According to the solution, a neural network is trained to determine an association degree among a group of words in a source or target language. The neural network can be used for translation verification and correction. According to the solution, a group of words in a source language and translations of the group of words in a target language are obtained. An association degree among the group of words and an association degree among the translations can be determined by using the trained neural network. Then, whether there is a wrong translation can be determined based on the association degrees. In some embodiments, corresponding methods, systems and computer program products are provided.