Hybrid Translation Platform with Crowdsourced Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine translation often generates numerous errors when translating text into multiple languages, resulting in an unprofessional look and feel for applications intended for a global audience.
Innovation Solution
A hybrid translation platform that combines machine translation with crowdsourced human correction, where machine translation is initially applied, and users can submit corrections through a graphic user interface, with verification processes involving human review or reverse machine translation to ensure accuracy, and updates are applied to all end-clients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine translation is used to translate text into multiple languages, then translation efficiency is improved, but translation accuracy deteriorates
Solution Approach 1:
The patent combines machine translation with crowdsourced human translation into a hybrid system. Machine translation provides initial translations efficiently, while human translators review and correct them, achieving both high productivity and high accuracy through the integration of automated and human processes.
Solution Approach 2:
The system implements a feedback mechanism where human translators review machine translation outputs, provide corrections, and these corrections are fed back to improve future machine translations. This continuous feedback loop enables the system to maintain high translation efficiency while progressively improving translation accuracy.
2Measurement precision
If crowdsourced human translation is used to improve translation accuracy, then translation quality is improved, but system complexity increases
Solution Approach 1:
The patent introduces a platform as an intermediary that coordinates between machine translation systems and human translators. This intermediary manages the workflow, assigns translation tasks, collects corrections, and integrates feedback, thereby organizing the complex crowdsourced translation process into a manageable system structure.
Solution Approach 2:
The translation platform serves multiple functions: it performs initial machine translation, manages crowdsourced human translation tasks, collects and validates corrections, and applies updated translations across multiple applications. This multi-functionality consolidates various operations into a single system, managing complexity through universal design.
3Reliability
If manual review and verification processes are implemented, then translation reliability is improved, but processing time increases
Solution Approach 1:
The system performs preliminary machine translation before human review, so that human translators only need to review and correct rather than translate from scratch. This preliminary action by the machine translation system reduces the time required for human review while maintaining high translation reliability through subsequent human verification.
Solution Approach 2:
The system implements selective human review where human translators focus on reviewing and correcting machine translation outputs rather than translating all content manually. This partial action approach maintains high reliability for critical translations while reducing overall processing time by leveraging machine translation for routine content.
Data Source
AI summary
Disclosed herein is a translation platform making use of both machine translation and crowd sourced manual translation. Translation is performed on pages in an application. Manual translations are applied immediately to local versions of the client application and are either human reviewed or reverse machine translated and compared against the original text. Once verified, the translations are applied to all end-clients.