Language Model Translation Prompts With Targeted Dictionary Terms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine translation systems using language models do not consider the overall meaning of a source text, leading to inconsistent translations and excessive computational processing due to unnecessary data input from bilingual dictionaries.
Innovation Solution
An information processing system that extracts relevant translation pairs and includes specific translation instructions to guide the language model, reducing the input to only necessary data and optimizing the translation process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all entries from the bilingual dictionary are included in the prompt, then translation consistency is improved, but computational processing time and load increase excessively
Solution Approach 1:
The patent extracts only the necessary translation pairs from the bilingual dictionary that are actually present in the source text, rather than including all dictionary entries. This selective extraction reduces the prompt size and computational load while maintaining translation consistency for the relevant terms.
Solution Approach 2:
The patent segments the bilingual dictionary into relevant and irrelevant portions based on the source text content. Only the segment containing translation pairs for terms actually appearing in the source text is included in the prompt, separating useful information from unnecessary data.
2Measurement precision
If all entries from the bilingual dictionary are input into the language model, then translation accuracy is improved, but device complexity and data processing requirements increase
Solution Approach 1:
The system extracts only the necessary translation pairs from the bilingual dictionary that correspond to terms actually present in the source text. This reduces the data volume and processing complexity while maintaining translation accuracy for the relevant terminology.
Solution Approach 2:
Instead of providing the complete bilingual dictionary (excessive action), the system provides only the partial set of translation pairs that are actually needed for the current source text, optimizing the balance between accuracy and complexity.
Data Source
AI summary
One or more processors extract, from a bilingual dictionary that includes pre-translation terms described in a first language and post-translation terms described in a second language respectively corresponding to the pre-translation terms, one or more pre-translation terms relevant to one or more source texts and one or more post-translation terms respectively corresponding to the one or more pre-translation terms. The processors input, into a language model, a translation prompt that includes the one or more source texts, the one or more pre-translation terms, the one or more post-translation terms, and a translation instruction that includes an instruction for the language model to translate the one or more pre-translation terms to the one or more post-translation terms. The processors acquire, from the language model, a translated text in which the one or more source texts are translated into the second language according to the translation instruction.