Discourse-Level Text Translation Using Contextual Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing discourse-level text translation methods fail to accurately translate texts due to lack of consideration for contextual information, leading to reduced translation accuracy and user experience.
Innovation Solution
A method and apparatus that translate discourse-level texts by obtaining relevant text units, including preceding and succeeding source texts, and preceding target texts, and using an encoding-and-decoding model to process semantic information, thereby improving translation accuracy by considering contextual relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the to-be-translated discourse-level text as a whole is directly translated, then the translation process is simple, but the accuracy of the translation result is not high
Solution Approach 1:
The patent segments the discourse-level text into multiple sentence units and translates them sequentially rather than translating the entire text at once. Each sentence unit is translated individually while maintaining access to previously translated sentences, which breaks down the complex translation process into manageable segments and improves accuracy without overwhelming computational complexity
Solution Approach 2:
The patent performs preliminary encoding of the entire source text before translation to capture contextual relationships. This preliminary action allows the translation model to access contextual information from preceding and succeeding sentences during the translation process, improving accuracy while maintaining an organized workflow
2Measurement precision
If contextual information from preceding and succeeding text is incorporated, then translation accuracy is improved, but the complexity of the translation system increases
Solution Approach 1:
The patent merges the source text encoding and target text generation into a unified encoder-decoder model that processes both source and target sequences simultaneously. This combining of functions within a single model architecture reduces system complexity compared to using separate models for each function while still capturing contextual relationships
Solution Approach 2:
The patent introduces an intermediate encoded representation of the source text that serves as a mediator between the input source text and the output target text. This intermediate representation captures contextual information from the entire source text and makes it accessible during translation without requiring direct complex interactions between all text elements
Data Source
AI summary
A discourse-level text translation method and device, the method comprising: acquiring a text to be translated, the text to be translated being a unit text in a discourse-level text to be translated (S101); acquiring an associated text of the text to be translated, the associated text including at least one of a preceding source text, a following source text, and a preceding target text (S102); and translating, according to the associated text, the text to be translated (S103).