E.g., 04/05/2020
E.g., 04/05/2020

Memsource Releases New Version of Machine Translation Quality Estimation Feature


Tuesday, 12 February, 2019

Memsource, the translation management system for global companies, has released the latest version of its second AI-powered feature, Machine Translation Quality Estimation (MTQE). The new version of this unique feature, which assigns quality scores to raw machine translation output, is based on redesigned AI models with a new neural network architecture. Initial testing indicates that, for certain language pairs, quality scores are available in up to 60% of segments, which could equate to up to 10% savings on post-editing costs.

“The new version of MTQE is an important addition to our AI solutions,” said Dalibor Frivaldsky, Memsource CTO. “With the new approach we’ve taken, MTQE has the potential to become the missing link in large-scale, seamless MT adoption.”

The changes to the second version include improving the segment coverage by up to four times and altering the scoring categories. Read more about how the Memsource AI team developed MTQE. The new categories make it easier for linguists to determine the level of post-editing required and identify what should just be translated from scratch. Find the updated scoring categories and descriptions at memsource.com/mtqe.

This update is the first in a series of AI-powered improvements we are making to machine translation management in Memsource. Stay tuned for further announcements in the coming weeks.

About Memsource

Memsource is the translation management system for global companies wanting to improve localization efficiency. 400+ languages, 50+ file types, 30+ MT engines, REST API, and patented AI make Memsource the TMS used by many of the world’s leading brands to reduce costs, automate workflows, and optimize the entire translation process.