Translation System Using Core Word Image Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional translation technologies face challenges in accurately translating text into a target language, as users struggle to determine the accuracy of translations, especially when the original language is not known, and errors can accumulate during translation and reverse translation, leading to low translation accuracy.
Innovation Solution
A server and user terminal system that extracts core words from translated text and provides corresponding image information, allowing users to correct translations by selecting alternative images, which correspond to homonyms or similarly pronounced words, thereby enhancing translation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional translation methods are used to translate text into a target language, then translation can be performed, but users cannot determine whether the translation is accurate when they do not know the target language
Solution Approach 1:
The patent introduces image information as an intermediary element between the translated text and the user. When a user selects image information corresponding to a core word, the system performs reverse translation to generate reference translated text, allowing the user to verify translation accuracy through visual comparison without needing to know the target language
Solution Approach 2:
The patent segments the translation verification process into discrete steps: selecting image information for specific core words, performing reverse translation for those selected words, and comparing the reference translated text with the original translated text. This segmentation allows users to verify translations systematically rather than attempting to evaluate the entire translation at once
2Measurement precision
If reverse translation is used to allow users to check translation accuracy, then users can verify translations, but translation errors accumulate during translation and reverse translation
Solution Approach 1:
The patent implements partial reverse translation, performing reverse translation only for selected core words rather than translating the entire translated text back to the source language. This partial action reduces the accumulation of translation errors while still providing sufficient verification capability for critical words
Solution Approach 2:
The system provides feedback by displaying reference translated text for selected core words alongside the original translated text, allowing users to identify discrepancies and request corrections. This feedback mechanism enables iterative improvement of translation accuracy without requiring complete reverse translation
3Measurement precision
If image information is provided for each core word to enable correction, then translation accuracy improves, but system complexity increases
Solution Approach 1:
The patent leverages the multi-functionality of existing image information and reverse translation capabilities already present in the system. Rather than creating entirely new components, the system repurposes existing image databases and translation engines to serve the additional function of translation verification and correction
Solution Approach 2:
The system performs preliminary actions by pre-associated image information with core words in the translated text. This preliminary preparation allows users to quickly select and verify translations without requiring the system to generate or search for image information in real-time, reducing operational complexity
Data Source
AI summary
A server, a user terminal, and a method for controlling the server and the user terminal are provided. The server controlling method includes receiving a text from a user terminal, translating the received text to generate a translated text, extracting at least one core word from the translated text, obtaining image information corresponding to the at least one core word with respect to each of the at least one core word, and transmitting the translated text and the extracted image information to the user terminal.