Precision at Scale: Navigating High-Velocity Quality Control for AI Translation
- Tutorial
- Terminologie, Übersetzung & Sprachtechnologien
- 12. November
- 11:30 AM (MEZ) - 12:30 PM (MEZ)
- C4.1
-
Simone Lamont
- Lionbridge
Inhalt
As global enterprises integrate Large Language Models (LLMs) into localization, the priority has shifted from simple cost-savings to industrial-grade accuracy and control. This session explores a rigorous, data-driven framework for reducing human-in-the-loop dependencies without sacrificing linguistic integrity. By using the SAE J2450 Translation Quality Metric as a foundational standard, we demonstrate how to validate legacy content and translation memory (TM) at scale. We will discuss the deployment of custom AI prompts and AI-enhanced Automatic Post-Editing (APE) to lock high-quality segments, reducing human review effort within months, not years. AutoLQA then continuously verifies that quality targets are being met at volume, closing the loop on automated validation. Attendees will leave with a concrete framework for moving from manual, sampling-based review to a high-velocity automated validation model, one that delivers the accuracy and control that global enterprises demand.
Das lernen Sie
- SAEJ2450, a control mechanism making high-volume, low-HITL pipelines trustworthy/auditable
- TM validation, APE, and AutoLQA enable rapid/measurable HILT reduction
- Automation built on quality standards deliver sustainable scale