0

Ive got some french text data (20000 lines) and corresponding english translation.

This data is specific to insurance domain and hence will have some restrictions due to vocabulary. Now when Im using google to translate to do it, its giving me 90% accuracy wrt translation. I want to improve this to 95+.

Is there a way/method wherein I can improve the translation output which Im getting by google by passing it through another algorithm based on insurance domain etc to improve upon the final accuracy?

vinita
  • 595
  • 1
  • 9
  • 24
  • 1
    Before you try this, you should make sure you have a way of verifying that you are actually domain-adapting the system in a general way, not just overfitting towards your (quite small) dataset. Eg. have more than one reference translation per unit or manually check the difference in output. – lenz Aug 17 '17 at 09:17
  • @lenz thanks , thats on my mind. But do you ahev any suggestion on how to implement another translation function on the output ? – vinita Aug 17 '17 at 10:15
  • I don't know the Google API because I've never used it, but maybe someone who does can help you. Why don't you retag your post with the `google-api` tag so you attract the right people's attention? – lenz Aug 17 '17 at 13:39

0 Answers0