Re: [GNU-linux-libre] nltk-data

[Sam Geeraerts]

Nicolás Reynolds wrote:
> Hi,
>
> this package, nltk-data, got caught on our filters due to a license that says:
>
>     Parts of NTLK-Data are distributed under various licenses,
>     as documented in their respective README files.
>     See: /usr/share/nltk/data/corpora/ -- [0]
>
> so we went checking those licenses, to find out that the package is composed by
> a bunch of data sets with very different licensing, very few of them free.
>
> the licenses include copyright plus redistribution allowed, cc-nc, bsd-like,
> only one gpl, and several doesn't even have a license, just the author name.
> though, these are collections of words... i don't know if copyright applies to
> them.

I picked a random package [1]. The license attribute says "May be used 
for non-commercial purposes.". The README in the zip says 
"redistribution permitted". The non-commercial clause means that it's 
non-free. It looks like structured data to me, so copyrightable.

> also, nltk (the software) has a method for downloading this data directly from
> the website... this qualifies as recommending unfree stuff, right?

From the description on the website [2] I'd say yes, non-free.

[1] 
http://nltk.googlecode.com/svn/trunk/nltk_data/packages/corpora/brown.xml
[2] http://www.nltk.org/data