16
$\begingroup$

I am looking for large (>1000) text corpus to download. Preferably with world news or some kind of reports. I have only found one with patents. Any suggestions?

Improve this question
$\endgroup$
4
  • $\begingroup$ This thread appears to be off topic. See meta.stats.stackexchange.com/questions/1032/…. $\endgroup$
    – whuber ♦
    Commented Jun 7, 2012 at 21:39
  • $\begingroup$ This question appears to be off-topic because it is about finding a data set, rather than doing statistical analysis $\endgroup$
    – Peter Flom
    Commented Nov 7, 2013 at 13:12
  • 2
    $\begingroup$ Well that's awkward, because this Q&A is really useful. $\endgroup$
    – Sideshow Bob
    Commented Jan 7, 2014 at 15:35
  • $\begingroup$ @guaka, please do not bump such old posts for such minor edits, especially a post that is closed. It is true that our style preference is not to have "thanks", but for something this minor, we'd just leave it. $\endgroup$
    – gung - Reinstate Monica
    Commented Mar 15, 2019 at 13:48

6 Answers 6

Reset to default
9
$\begingroup$

Do not the Wikileaks texts suit you?

Improve this answer
$\endgroup$
1
  • $\begingroup$ But how could I download them in .txt $\endgroup$
    – Dimitar Vouldjeff
    Commented Nov 24, 2011 at 22:44
6
$\begingroup$

What about wikinews? Here's the latest database dump I could find: http://dumps.wikimedia.org/enwikinews/20111120/

You probably want the "All pages, current versions only."-version.

Improve this answer
$\endgroup$
2
  • $\begingroup$ This no longer works. $\endgroup$
    – vy32
    Commented Feb 6, 2016 at 21:48
  • $\begingroup$ dump link no longer works. dataset by region is small and outdated $\endgroup$
    – HappyCoding
    Commented May 24, 2016 at 8:38
6
$\begingroup$

The reuters text corpus is a classic in the field, and can be found here

Improve this answer
$\endgroup$
2
  • $\begingroup$ It's not the most interesting (or diverse) corpus. The license is also restrictive relative to Wikileaks (public domain US documents) or wikinews. $\endgroup$
    – ariddell
    Commented May 17, 2013 at 14:57
  • $\begingroup$ @ariddell i agree, but it is commonly used in introductory NLP examples, and its large enough to be useful in learning but small enough to be analysed on a good laptop. $\endgroup$
    – richiemorrisroe
    Commented May 28, 2013 at 10:50
3
$\begingroup$

http://endb-consolidated.aihit.com/datasets.htm contains 10K companies with textual descriptions

Improve this answer
$\endgroup$
1
  • $\begingroup$ is currently outdated $\endgroup$
    – Quonux
    Commented Mar 25, 2017 at 22:12
1
$\begingroup$

If recency is not an issue, you can try

http://www.infochimps.com/datasets/20-newsgroups-dataset-de-duped-version

and there are other many more similar dataset in infochimp depending on your budget.

Regards, Andy.

Improve this answer
$\endgroup$
1
  • $\begingroup$ This no longer works $\endgroup$
    – vy32
    Commented Feb 6, 2016 at 21:48
1
$\begingroup$

If you want precomputed n-grams, you could try the google books archive:

http://books.google.com/ngrams/datasets

Improve this answer
$\endgroup$
1
  • $\begingroup$ how this one can be used? $\endgroup$
    – HappyCoding
    Commented May 24, 2016 at 8:39

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