JSTOR and false DOI Numbers

DWL-SDCA
With citations from the American Journal of Sociology, the doi number that JSTOR provides isn't the real doi. The metadata for this journal (and probably others) is an internal JSTOR system identifier information that is interpreted as a doi number. The JSTOR internal tracking numbers are in exactly the same format as a real doi. Using these JSTOR numbers in a http://dx.doi/ URL leads to a doi error screen.

Now, if only there was a way to use Zotero to get the true doi numbers from the University of Chicago Press site in a batch process...
  • noksagt
    This was reported in the dev list the other week:
    http://groups.google.com/group/zotero-dev/browse_thread/thread/16c9ef2e44c72aef?pli=1

    I haven't looked into it at all, but useful questions: are the registered JSTOR DOIs denoted somehow or does removing bad DOIs require removing all DOIs? What are long-term plans for registration of more DOIs?
  • DWL-SDCA
    I don't know the specific answers to your questions. However, I talked with a couple of people on the JSTOR staff and was told that although the numbers cosmetically resemble doi numbers, they are not and will not be registered as doi numbers. In the case of the American Journal of Sociology the real doi numbers must be obtained from the journal website. The JSTOR internal tracking numbers are useless to anyone outside of the JSTOR staff.
  • dunning
    Could somebody look into this again? It seems that Zotero is still exhibiting the behaviour described above.
  • aurimas
    This should be fixed now. You can update your translators via Preferences -> General -> Update Now.

    This was actually fixed before, but it looks like JSTOR itself is now exporting the fake DOIs as real DOIs in their RIS export format.
  • dunning
    Many thanks! I did notice that, but I wasn't sure whether this was the way it had been done in 2009. Just for future reference, can Zotero distinguish between real and fake DOIs, or does it simply not take DOIs from JSTOR at all?
  • aurimas
    There is nothing particular about a DOI that makes it real or fake (e.g. unlike ISBN numbers, which have check digits and can be easily validated). However, "real" DOIs, in order to be useful (this applies to ISBNs too), have to be registered with a DOI registration authority (CrossRef being one of them). Thus, there is no way for Zotero to immediately know if a DOI is "real" or not. Every time you import an article from JSTOR, Zotero queries CrossRef to check if the DOI has been registered and if it has, it will add the DOI to the metadata.
  • dunning
    Very good point. That seems very sensible; I'm rather amazed that you managed to make that change in a span of twenty minutes. Thanks again!
  • aurimas
    To be fair, we were already doing this (the code was already there), there was just a bug due to an assumption that DOIs were not being exported in RIS format (which apparently has changed). So this was a fairly easy fix. Thanks for reporting!
  • DWL-SDCA
    Does this DOI / CrossRef check only apply to JSTOR? Some publishers release their ePub ahead-of-print articles and send metadata to indexing services long before the DOIs are registered with CrossRef.
  • aurimas
    edited October 10, 2013
    Yes, this check is only performed for JSTOR. We could perform a similar check for all DOIs using the CrossRef RA look-up tool (e.g. http://doi.crossref.org/doiRA/10.6220/joq.2012.19(1).01 ), but in general I don't think this should be necessary. DOIs should not be used published without being registered. JSTOR is (or at least was) a bit of an exception here, because they were not supplying the DOI and we were actually trying to construct the DOI from their internal identifiers.
Sign In or Register to comment.