Requests for dumps

← Requests and proposals

Requests for dumps

Archives →

Ask on this page for dumps that you need but that are not available from download.wikimedia.org, or that you cannot reasonably download from that site because of bandwidth limitations (on your end or on Wikimedia's end). Please specify which wikis you want dumps for and what data you want in the dump. If you want dumps shipped to you on some type of media, please note the kind of media, and provide a way for someone to get in touch with you to coordinate shipping (you will have to pay for materials and shipping).

If you want an unusual set of data in a dump, please provide the query necessary to complete your request - it'll be slower if the developer concerned has to write the query.

Dump of Romanian Wikipedia articles

I am an Natural Language Processing researcher, and need to have access to a dump of Romanian Wikipedia. I would prefer to have only the political articles, however the entire set of Romanian articles will do as well. The last dump that was performed had 25 errors, and the file pages-articles.xml.bz2 was not created. I would greatly appreciate your support. 11:00, 16 February 2006

Full history dump of all languages (from late 2006)

Full-history dumps of all languages are large. Some are too large for the dumps to complete successfully. The last full-history dump of en:wp that works at the moment is from mid-August (another is underway; let's hope this one completes without trouble...) I would like a copy of the latest full-history dump of each language of each project on a hard disk; along with a summary list of the dates of each dump. If you have such data (dates don't have to be the same for each dump) and want to help, let me know; and I will send you a drive and return packaging. See also below. +sj | help with translation |+ 03:37, 1 November 2006 (UTC)

Image dumps (from November 2005)

I would like image dumps with metadata; for instance the Wikimedia image dumps from November 2005 from each project and from commons. More recent dumps would be excellent; but I don't know that any have been made.

The commons dump alone is 290GB. It will be cheaper and faster, and less strain on the dumps server, for me to ship a 500G disk to someone who could copy the dump. If you have access to a dump and could make me a copy, please get in touch by email to work out the details. +sj | help with translation |+ 03:20, 1 November 2006 (UTC)

Make random-subset dumps available from downloads.wikimedia

Many researchers just want a randomized subset of articles, or users, or revisions to run tests and experiments and programs on. At the moment, they either have to screenscrape what they want, or download the entire dump of the project they are interested in -- for en:wp, this is difficult to do regularly or efficiently.

Please consider adding subsets of en:wp and other wikis to the default collection of dumps: for instance, all revisions and users and metadata associated with 1000 random articles; or all metadata and revisions of articles touched by 1000 random users, along with their userspace pages. Here randomness could mean almost-even distribution across the set of all users with at least one edit, or articles that are not blank or redirects. Deciding how to randomize and what to select is not trivial; if you are interested in doing this, please ping the wiki-research-l list for more specific details on what good random subsets would include. +sj | help with translation |+