Slashdot Mirror

I think you'll find that's exactly how they're testing it... Here's some sample test data...

http://matrix.statmt.org/matri...

It's not perfect, but it actually looks pretty good.

I'm just wondering how you reached this conclusion? Did you read the article and the linked papers? Or did you just skip all of that and go straight to trolling?

I think you'll find that's exactly how they're testing it... Here's some sample test data...

http://matrix.statmt.org/matri...

It's not perfect but it actually doesn't look too bad.

I'm just wondering how you reached this conclusion? Did you read the article and the linked papers? Or did you just skip all of that and go straight to trolling?

I think you'll find that's exactly what they're doing here some sample tests ...

http://matrix.statmt.org/matri...

So you read the article and the linked papers and came to this conclusion? Or did you just skip all of that and go straight to trolling?

The answers to most of your questions are available, you just need to follow the references. They didn't discuss those details because it wasn't relevant to the paper's main topic whether they were using Mandarin or Cantonese. Besides, they didn't create the training data. They used an existing data set someone else had created, which is actually a collection of several data sets from different sources. So if you want all the details, here's the data.

Today's most successful approaches use statistical (or hybrid) machine translation. Moses is a great open source example that presents a fine baseline when training with good parallel language data and language models. However, it uses Bayesian statistics that uses phrase co-occurrences and words in context to define a probabilistic generative model for translation. While it provides impressive results, it is still a far cry from fluent (and adequate) output. Additionally, if he wants to claim that speech translation will also be up to par, he is also requiring researchers to perfect speech recognition. Granted, he's giving us researchers 18 years to develop a solution, I still think he's underestimating the challenges statistical systems need to overcome.

If you're interested in learning more about machine translation, I invite you to take a look at the EuroMatrixPlus project.

There is an academic statistical machine translation system: http://demo.statmt.org/index.php This is open source. Help improve it!

State-of-the-art is Moses for decoding with Giza++ for word-alignment for training. The MT domain has an egyptian naming tradition for some reason (Moses is the open-source successor to Pharaoh). OG.