6 ms·
> Unfortunately, I am not aware of any publicly available dataset of (mostly) grammatically correct English. How about books?
by topynate 10y ago
> Unfortunately, I am not aware of any publicly available dataset of (mostly) grammatically correct English.
How about books?
- wiz21c 10y agoGuess who's got that massive amount of data nowadays ? Google, facebook and co. So good luck if you want to work on AI stuff...
- mortehu 10y agoAnd Project Gutenberg: https://www.gutenberg.org/ https://www.gutenberg.org/ The language in many of those books may be a bit archaic, though. See also Distributed Proofreaders: http://www.pgdp.net/c/ http://www.pgdp.net/c/
- jcoffland 10y agoThere are a fair number of OCR mistakes in many of the Gutenberg texts. It would be interesting to try to correct them. They are however not the same types of errors which are addressed here.
- rspeer 10y agoWho's got that massive amount of data nowadays? Everybody who wants it. I recommend http://opus.lingfil.uu.se/ http://opus.lingfil.uu.se/ as a starting point to find lots of sources of quality, multilingual data. And if all you want is sheer quantity, at rather the expense of quality, then there's a several terabyte Web crawl at http://commoncrawl.org/ http://commoncrawl.org/.
- raverbashing 10y agoWikipedia, Books are usually ok Online forums are a disaster (HN is not too bad, but there are a lot of natives and non-natives fumbling some things)