Quality assessment of the Autodesk productivity test

To ensure the use of machine translation does not lead to a loss of final translation quality, we carried out a blind sample quality assessment of both the human-translated and post-edited MT versions. The quality assessments were made by Autodesk's usual, highly experienced linguistic quality assurance team, using Autodesk's translation quality criteria.

Quality assessment

Beyond the chart, the outcome of the quality assessment showed the following:

  • Reviewers are consistently unable to tell the difference between HT and post-edited MT.
  • There is no pattern that suggests a loss of quality induced through the use of MT.
  • The two Italian outlyers are only coincidental.

Back to intro