Matching Data White Paper

A new technique to optimize data selection for machine translation training

Matching-Data-White-Paper-January-2019-1In this full version you will find all aspects that led to the creation of TAUS Matching Data Service.

Laying out the history of of language data sourcing as well as techniques and challenges associated, this white paper introduces:

  • Matching Data as a solution to the constraints described. Matching Data is a high-performance clustered search methodology, based on data selection techniques developed in the DatAptor project.