 |
CiteULike |  |
|
 Search
|
|
Register | |
Log in |  |
I just fixed a bunch of my articles that didn't have publication years, page numbers or other random details. A big thumbs-up to this feature!
Can I be confident that the metadata from Crossref are correct? Do they have some efficient checking process within Crossref? Where do they get their metadata from? I am asking because one of the articles I updated has several collaborating authors but the crossref update scratches out all the authors except the first one.
If Crossref data is dependable, then an option to update entire libraries (or sets of tags) via crossref might be even more useful. Or Citeulike could give users an option to import metadata from crossref instead of the source while posting the article.
Posted by Zephyrus on 2009-05-27 14:53:41.
4 replies. Login or join this group to post to this thread.
Can I be confident that the metadata from Crossref are correct?
If Crossref data is indeed quite superior to any of the other sources
CrossRef get their metadata from the publishers. In theory, CrossRef data should be better than that on the publishers site, if only because it's in a standard format. In practise the situation is a lot more complicated. Sometimes the publishers will send bad data to CrossRef. Sometimes late. Some publishers are luddites and still don't get the whole interweb thing and won't put sufficient resources into it. etc. etc.
Actually, we've only recently been given permission to use CrossRef in a batch or heavy duty way, so we're looking at ways to use CrossRef data to augment the publisher's data. One problem is that each lookup slows things down, so an import (BibTeX, say) of 1000 records will takes AGES. Perhaps we can make the upload asynchronous but one of the advantages of the net is that errors are immediate so we'd then have to find a way to alert the user if a particular batch/record failed. Yuck.
Posted by thegoose on 2009-05-27 15:11:43.
Wouldn't it be awesome if you could compile a bibliography using citeulike, feed the output .bib file through crossref, have it check if all the details are fine and have suggestions/corrections from crossref presented to you in the form of a "corrected" .bib file?
Or you could have a button during export that automates this process and then churns out a "corrected" export option. Is such a thing possible?
PS: I love the fact that I can correct all the "aop" fields in my citations with crossref. They are big pain that never seem to get updated on NCBI and publisher pages.
Posted by Zephyrus on 2009-06-10 14:35:46.
As I mentioned above, crossref's data (especially authors) isn't anywhere near as reliable as it should be. For example, there's this hardly heard of paper:
http://www.citeulike.org/user/thegoose/article/197238
I dunno, 100 authors? Crossref has, er, 1. This is quite common. I really think an automated crossref "corrected" export would do as much harm as good - I feel it really needs the human touch unless we can develop some heuristics to decide when CrossRef's data is better than what we've got through the normal channels.
Posted by thegoose on 2009-06-10 22:23:40.
Oh well. You know what they say about human touches - prone to random illogical errors. ;-)
Posted by Zephyrus on 2009-06-11 02:22:20.