3

Are European Union parallel multilingual texts - regulations, directives, especially the debates of European parliament - ideal for machine learning of machine translation, e.g. with neural networks? My guess is that they are ideal, but I have not seen they to be used in actual research papers. If not, then - why they can not be ideal?

I am specifically interested in the grammar induction as the by-product of the machine translation learning a la https://arxiv.org/abs/1805.10850 .

4

Europarl is a classic corpus for research papers, used at the main conference - WMT - and by some of the top people in the field.

enter image description here

It would be useful for training a translation system specifically for European parliament domain.

But Europarl, like any domain-specific corpus, is not ideal for training a production-strength open-domain machine translation system.

How many times do the top queries like how r u or ai eu se te pego in the corpus? To say nothing of laham taz-ziemel or gradient descent.

| improve this answer | |
  • Now that I’ve googled what laħam taż-żiemel means, I think it may very well appear in some EU regulation, just like many other food items do. – Emil Jeřábek Sep 26 '18 at 12:04
  • You can Google site:europarl.europa.eu "laham taz-ziemel"`, no results. If you replace with with *parmesan or cognac you will see get results. – Adam Bittlingmayer Sep 26 '18 at 15:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.