CiteULike datasets | |
---|---|
Type |
Dataset |
Link |
|
Source |
Ckan.net |
From the href="http://www.citeulike.org/faq/data.adp">data page:
[HTML_REMOVED]Who-posted-what data[HTML_REMOVED]
The latest data snapshot can always be downloaded at
href="http://static.citeulike.org/data/current.bz2" target="_blank" rel="nofollow">http://static.citeulike.org/data/current.bz2
Older datasets are available on a daily basis and can be found
at URLs of the form href="http://static.citeulike.org/data/2007-05-30.bz2" target="_blank" rel="nofollow">http://static.citeulike.org/data/2007-05-30.bz2
Data is available from [HTML_REMOVED]2007-05-30[HTML_REMOVED] onwards.
The file constitutes an anonymous dump of [HTML_REMOVED]who[HTML_REMOVED] posted
[HTML_REMOVED]what[HTML_REMOVED] and [HTML_REMOVED]when[HTML_REMOVED] the posting took place. There is no data in this file which is not already available publicly through the web site, so there are no privacy implications for making it available. The advantage is that it's available in one file rather than having to spider the entire site to get at the information (please don't do that!).
The file is a simple unix ("n" line endings) text file with pipe ("|")
delimiters. The columns are:
[HTML_REMOVED]The CiteULike article id which was posted[HTML_REMOVED]
[HTML_REMOVED]An obfuscated representation of the username (a salted MD5 hash of
the true username). Again, it is possible to piece back together what the true username is by scraping the site, but I'd rather you didn't do that. The reason I've gone to the trouble of obfuscation is primarily a slightly paranoid anti-spam measure[HTML_REMOVED]
[HTML_REMOVED]The date and time the article was posted to the site[HTML_REMOVED]
[HTML_REMOVED]The tag the user used to post it[HTML_REMOVED]
[HTML_REMOVED]NB[HTML_REMOVED] Note that if a user posts an article with [HTML_REMOVED]n[HTML_REMOVED]
tags, then this will result in [HTML_REMOVED]n[HTML_REMOVED] rows in the file
[HTML_REMOVED]Article linkout data[HTML_REMOVED]
Mapping CiteULike article_ids to resources on the web can be done
with the linkout table. The current snapshot is available at href="http://static.citeulike.org/data/linkouts.bz2" target="_blank" rel="nofollow">http://static.citeulike.org/data/linkouts.bz2
Openness: OPEN (?)
License: no license specified but manner in which it is made available suggests it is open. Access: good. bulk: yes.