-
Notifications
You must be signed in to change notification settings - Fork 7
RelatedWork
There's been some interesting related work in using crowdsourcing-like techniques to build corpora.
Here's what we know about so far:
http://www.cs.umd.edu/hcil/monotrans/
"... an iterative protocol in which monolingual human participants work together to improve imperfect machine translations."
http://www.cs.cmu.edu/~vamshi/Vamshi/Publications.html
"Active Learning for Machine Translation in Scarce Data Scenarios" -- using active learning with Mechanical Turk to bootstrap an en-es corpus for training SMT.
http://wiki.apertium.org/wiki/Tradubi
(project seems to have died, though)
This looks awesome. If we could just bring up an instance of this, that would be close to the right thing.
Really nice site for collecting example sentences translated into a variety of languages.
"the open parallel corpus".
From MSR. Seems really relevant. They're about getting the content from English into smaller wikipedias. But only English for some reason? Relates to Bing Translator somehow. Project seems dormant.