Advanced Features to fine-tune Machine Translation output

In order to improve the Machine Translation output, Intento offers multiple fine-tuning features, such as:

• Content Moderation

  Add content moderation (i.e., profanity filtering) in the translation pipeline to check MT output for profane words, hate speech, racism, etc., and act according to the company's policy (return a note with translation, mark, or delete profanity, etc.).

• Same Language Detection

  Exclude parts of the source text from translation requests when the detected source language is the same as the target language. This reduces the cost of translation.
  English:
  It s like a famous German song says: Neunundneunzig Luftballons Auf ihrem Weg zum Horizont
  German:
  Wie ein berühmtes deutsches Lied sagt: Neunundneunzig Luftballons Auf ihrem Weg zum Horizont

• Non-translatable Markup

  Use specific markups in translation requests to highlight any content that should remain unchanged in translations.
  

  [[non translatable text]]
  English:
  All products are marked with "[[Made in the USA]]".
  German:
  Alle Produkte sind mit "Made in the USA" gekennzeichnet.

• Linebreak Characters Handling

  Process linebreak characters in the source text more flexibly to increase the quality of MT output (i.e., ignore linebreaks or preserve them when necessary).
  

  English:
  All tools you need for this procedure <br />
  you can find it in the bottom tray.

  German:
  Alle Werkzeuge, die Sie für dieses Verfahren benötigen, <br />
  finden Sie im unteren Fach.

• Punctuation Correction

  Make fully configurable punctuation corrections in the source text and the translation to increase the output quality (i.e., remove or insert spaces before punctuation). Gender bias correction and original language improvement.
  English:
  This is what I call "a perfect vacation."
  French:
  C'est ce que j'appelle « des vacances parfaites ».

• Tone of Voice Control
  

  Set a specific tone of voice (formal/informal) for translations into relevant languages. This can also work with providers that do not natively support the tone of voice control.
  English:
  Please kindly reboot the printer.
  German (formal):
  Bitte starten Sie den Drucker neu.
  German (informal):
  Bitte starte den Drucker neu.

• Abbreviations
  

  Detect abbreviations in the source text according to the user's dictionary and expand them before sending for translation to make machine translations more precise.
  Dictionary:
  n8 = Nacht
  German:
  Gute n8.
  English:
  Good night.

• Redaction

  Detect personal data (i.e., names, addresses, bank details, etc.) in the source text with internal or third-party AI services and replace the data with dummy values to comply with data protection requirements.
  English:
  Hello. My name is Inigo Montoya. You killed my father. Prepare to die.
  English (pseudonymized):
  Hello. My name is John Smith. You killed my father. Prepare to die.