Phrase Table Combination Recording for Translation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Statistical machine translation systems face challenges in determining the specific combinations of phrase pairs used to generate candidate target texts, which hinders the updating of scores in the phrase table and thus affects translation accuracy.
Innovation Solution
A method is implemented to manage a server with a phrase table, where combinations of phrase pairs are recorded to identify the specific pairs used to generate candidate target texts, allowing for score adjustment based on user feedback to improve translation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If phrase-based statistical machine translation is performed using a phrase table, then translation capability is provided, but the ability to identify specific phrase pair combinations used to generate candidate target texts is insufficient, hindering score updates
Solution Approach 1:
The patent applies preliminary action by recording combination information in advance during the candidate target text generation process. The combination information recording unit stores which phrase pairs from the phrase table were used to generate each candidate target text before any evaluation or scoring updates occur. This ensures that when score updates are needed later, the system already has the necessary combination information readily available, eliminating the need to retrace the generation process and preventing information loss.
2Reliability
If translation accuracy is improved through score adjustment, then translation quality increases, but the system complexity increases due to additional recording and management mechanisms
Solution Approach 1:
The patent applies self-service by designing a system where the combination information recording unit automatically captures and stores the phrase pair combinations used during candidate target text generation without requiring external intervention. The system serves itself by maintaining its own training and evaluation data structures, where combination information is seamlessly integrated into the existing translation pipeline. This automatic self-recording mechanism reduces the need for complex external management systems while enabling accurate score adjustments for improved translation reliability.
Data Source
AI summary
Provided is a method of managing a server having stored thereon a phrase table containing phrase pairs each being a pair of a phrase in a first language and a phrase in a second language different from the first language. In the method, a processor receives combinations of the phrase pairs used to generate respective candidate target texts which are generated by translation of a source text in the first language into a target text in the second language as candidates for the target text in the second language, and records combination information (for example, a data structure and paths) in the server, the combination information identifying, in the phrase pairs contained in the phrase table, the combinations of the phrase pairs used to generate the respective candidate target texts.