Group: CiteULike-discussion - Forum Thread

Cases:

• I add a paper to my library and mark it to read later. Several weeks later I run across the same paper again. I have forgotten that I already added it to my Citeulike library. I add it again, not realizing that there are now two entries for the same paper in my library.

• I find a paper that I want to add. It's not from a supported website, so I enter the details of a paper manually. Sometime later I notice that the same paper is already in someone else's library in Citeulike and I could have saved time by copying to my own library. I also notice now that I have made several typos while doing the manual post.

• I am researching a new area and am unfamiliar with the seminal, well-known works of the field. I see there are statistics on Citeulike of how many people share a paper. I think it could be useful for me to find popular papers in my field. But I soon find out that there are many different entries for the same paper. So the statistics of the paper do not reflect the real number of people who have the same article in their library.

Feature suggestion: These problems could be solved if Citeulike used a single database entry for the static information of a paper (i.e. bibliographic data). A method like fuzzy-matching for the authors name and paper title could identify identical articles. A new article that matches one in the database would use the existing entry information. This way identical articles would not be posted twice to a library. The bibliographic data would also become more complete and accurate as several users would verify it as they added the paper to their personal library. Users could also save time during a manual post by first only entry the author and title, and checking whether it already exists in the Citeulike database. Finally, the statistics would accurately reflect the real number of people who hold the same paper.

Of course the individuals-specific data like notes, priority, tags, etc. could be stored separately from the paper's static information.

A shortcut to this might be to work with the doi. Admittedly coverage isn't universal, but it is a unique id associated with papers?

Thanks for your message.

We try very hard to only store one 'master' record for an article.

Here's an example: I searched Ingenta Connect for the term cancer. I selected the first article in the returned list - title is: Methodological issues in online data collection and posted it to CiteULike. Note that the URL of the article is something horrible, in my case this useless and non-permanent URL

When you've posted it to CiteULike, you'll see, when you go to that individual article in your library that there are two links to the online article:

• Ingenta Connect
• DOI

Note that both of these links are canonical ('proper' and permanent) links, not subject to the vagaries of links which were the result of the search. They have both been derived from looking at the article whose link was posted.

Now, if I go to Blackwell Synergy and find the same article (you can cheat and just use the DOI link on the article page if you like), you end up seeing this page.

If you now post that link to CiteULike, you will be taken straight to the article page for the the same article that you posted previously. Note that you will also see that the Online Article links have been augmented with the link direct to Blackwell Synergy:

• Ingenta Connect
• DOI
• Blackwell Synergy

This works for articles which are posted via plugins for supported sites. Does this answer your first and third points?

The second point, about manually posted articles, is more tricky -- we don't try to determine whether links point to the same content or not, and we don't try to match based on metadata. We intend to keep adding plugins for supported content. If you are reading a journal which isn't supported, please let us know and we'll try to write a plugin for it. Of course, you can always write your own plugins - details are here