Quality Evaluations + Component Testing

The integration of automated translation components, such as machine translation (MT), into the localization workflow yields great efficiencies as the quality of the MT system is monitored and tested. With the statistical analysis of the quality of MT output, you will be positioned to make informed choices from the many available open-source and off-the-shelf MT offerings. Simultaneous testing of applications in multiple domains allows for high-quality products in a fast-paced development environment.

Appen provides customized linguistic data and resources for the development, testing, and tuning of our clients’ language technologies. Services include test plan development and implementation, test data collection and verification, acceptance testing of third-party deliverables, and directed and undirected user testing.

The range of quality evaluations and component testing includes the following:

  • Machine translation testing and evaluation
  • Machine translation system improvement
  • Comparison and statistical analysis between machine translation systems
  • Data evaluation for training machine translation systems
  • Alignment of bilingual text data for training MT models
  • Error typology for MT output and high-level error reports
  • Testing of web and app contents for mobile phones or tablets
  • Quality evaluation of automated translation systems
  • Assessment of interactive voice response (IVR) applications, including recording the IVR application in the target country, transcription, and translation of the recording
  • Validation of text-to-speech (TTS) output for various domains