Machine Translation Apparatus Unifying Term Dictionary and Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine translation systems often produce inconsistent translations due to independent development of technical term dictionaries and translation memories, leading to variations in translated terms for the same technical term within a document.
Innovation Solution
A machine translation apparatus that includes a technical term dictionary, a translation memory, a corresponding translation pair extraction unit, and an unmatch check unit, which compares and checks translated terms from both sources to ensure uniformity by creating corresponding translation pairs and identifying unmatched pairs, allowing users to edit and unify translations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If technical term dictionary and translation memory are independently developed, then each can be optimized for its specific function, but translated terms become inconsistent between the two sources
Solution Approach 1:
The patent merges the technical term dictionary and translation memory into a unified translation apparatus that integrates both data sources. The corresponding translation pair extraction unit combines terms from the technical term dictionary with translation memory entries, and the unmatch check unit compares translated terms across both sources to ensure consistency, thereby resolving the contradiction between functional optimization and translation consistency.
Solution Approach 2:
The unmatch check unit provides feedback by detecting and reporting inconsistencies between translated terms from the technical term dictionary and translation memory. This feedback mechanism allows the system to identify and resolve translation inconsistencies, ensuring uniformity while maintaining the benefits of both independent data sources.
2Adaptability or versatility
If both technical term dictionary and translation memory are used for translation, then translation coverage is improved, but translated term uniformity deteriorates
Solution Approach 1:
The corresponding translation pair extraction unit acts as an intermediary that processes and compares translation information from both the technical term dictionary and translation memory. It extracts corresponding translation pairs and feeds them to the unmatch check unit, which ensures uniformity by identifying and resolving inconsistencies, thus maintaining translation coverage while improving term uniformity.
Solution Approach 2:
The unmatch check unit provides feedback on translation consistency by detecting mismatches between terms from different sources. This feedback enables the system to maintain both translation coverage and uniformity by identifying inconsistencies and prompting corrections or updates to ensure consistent terminology across all translation outputs.
Data Source
AI summary
A technical term dictionary stores terms of a first language and each translated term of a second language. A translation memory stores source sentences of the first language and each target sentence of the second language. A translation unit translates a text of the first language into an output text of the second language by the technical term dictionary and the translation memory. A corresponding translation pair extraction unit extracts, from a source sentence matched with a sentence of the text and a target sentence corresponding to the source sentence stored in the translation memory, a corresponding translation pair comprising a word in the source sentence and a translated word corresponding to the word in the target sentence. An unmatch check unit checks whether the translated word does not match with a translated term corresponding to the word in the output text.


