EBSCO translator does not save DOI

mronkko
I noticed that EBSCO translator either does not save DOIs or that this does not work for all articles.

Here is a test case

http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=39564295&site=ehost-live


For this particular item, also the date parser fails and the item list shows year as 0000
  • mronkko
    This item is also imported incorrectly. The date is not included.

    The item is recognised as report, but it is really a journal article, but this is probably an error in the database.

    http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=57366258&site=ehost-live
  • adamsmith
    I can confirm the DOI issue - I'll look into fixing that.

    Are you running the latest Zotero version? The date parser has been updated for 4.0 and it's saving this for me correctly.

    The second article is indeed in the database incorrectly - the RIS has
    TY - RPRT
    feel free to report that to EBSCO
  • aurimas
    I have a fix for DOI (will push that in a bit).
    The item is recognised as report, but it is really a journal article, but this is probably an error in the database.
    The problem is on their end.

    TY - RPRT
    M3 - Report

    But I'm open to ideas about fixing their metadata and detecting correct item type
  • mronkko
    Yes, this is with 4.0.3. The data parser does not seem to work.
  • adamsmith
    @Aurimas - I wouldn't try to hack the item type there. This really should be fixed at their end and I'm concerned that too many fixes will make translators ugly (and thus harder to maintain) and more fragile. mronkko - would you mind reporting this?
  • adamsmith
    odd - is this with Firefox or Standalone?
  • mronkko
    I will report it to EBSCO.

    I might have not been clear on my error report. Forget what I said about date before, I will explain this again.


    When I import an item from this URL

    http://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=39564295&site=ehost-live

    The date is imported as "Apr2009 Supplement 1". The Zotero date parser fails to parse the year. This is indicated by the item list showing "0000" in the year column for this item.

    This is not a translator issue, but the issue in Zotero date parser, so I should have probably started a new thread.
  • aurimas
    odd - is this with Firefox or Standalone?
    I can reproduce this on Firefox and I can see where this is done within EBSCOhost.
    The date is imported as "Apr2009 Supplement 1". The Zotero date parser fails to parse the year. This is indicated by the item list showing "0000" in the year column for this item.
    I'm not sure if Zotero needs to be able to parse a mess like that. This is all coming from a workaround that I added to improve date imports (I think), but I don't recall the test case right now. Perhaps we're just better off getting rid of the workaround (like adamsmith was saying)
  • aurimas
    Both fixes pushed. Update your translators and give it a go. That article is still imported as Report
  • mronkko
    The RIS field is

    Y1 - 2009/04/04/Apr2009 Supplement 1

    I think that that maps nicely to the zotero date format year/month/day/date-part

    (I assume that ebsco translator uses RIS import translator)
  • mronkko
    I deleted all the translators, restarted Zotero and tested importing. Both problems remain.

    Should I run translators from github directly?
  • aurimas
    The RIS field is
    Y1 - 2009/04/04/Apr2009 Supplement 1
    I think that that maps nicely to the zotero date format year/month/day/date-part
    The comment for the "workaround" was
    // If we have a double year in risDate, use last section
    Based on some of my other comments in the code, I think this may have been an improvement for a case like
    Y1 - 2009///Spring2009
    But I never noted the item, so I'm not sure if that's really the case.
    Should I run translators from github directly?
    Give it a few minutes. Translator updates should be instantaneous, but maybe it's taking a little longer today.
  • aurimas
    I deleted all the translators, restarted Zotero and tested importing. Both problems remain.
    Did you also update translators (Preferences... -> General -> Update Now)?

    You shouldn't need to delete translators or restart Zotero (just refresh the page) for translator updates (unless you have your reasons)
  • mronkko
    I did Update now first, but it did not help. So I deleted the translator and restarted to be sure.

    The EBSCO translator in now states.

    "lastUpdated": "2013-04-06 16:05:00"

    But both issues remain.
  • aurimas
    Are you using Firefox or Standalone? I verified that the updates went through correctly, so it's something specific to your setup.
  • mronkko
    Standalone and Safari connector. I tested with Firefox and it worked OK.

    I do not expect to import anything from EBSCO in the near future, so maybe the issue with Standalone will fix itself over time.
  • adamsmith
    Standalone does something more complicated with translator - like load them once a day or so - Simon explained that once, I'm sure Aurimas remembers that more precisely.
  • aurimas
    I assume when mronkko said he updated translators via "Update Now", he meant on Zotero Standalone.

    The discussion is here: https://groups.google.com/d/msg/zotero-dev/07GNRJ0qjso/T6GyfMsdP3cJ

    You should be able to force translator update from Zotero Connector Options -> Advanced. I'm not sure how you access that in Safari
  • mronkko
    I contacted the EBSCO support and they basically told that the item type comes from the publisher. In this case the item in question is marked as "Report" on their site.

    http://link.springer.com/journal/10551/98/3/page/1

    So it seems that this journal uses the terms "Report" and "OriginalPaper" for journal papers. My guess is that "Report" is like a research note. So the EBSCO data are correct here.

    But is there really a case where the EBSCO translator would need to produce a report item? Could all the items just be articles?
  • adamsmith
    EBSCO stores a huge variety of data - there are certainly newspaper and magazine articles as well as book chapters - I would be surprised if they didn't have genuine reports, too.

    Aurimas - how would you work around this?

    (We do get this correctly from Springer btw., but that's easier and we get the item type from the URL, not the metadata there).
  • aurimas
    My approach would be to check for something like JO tag to determine if it's a journal article. I'll have to look through the RIS spec to see what types make use of this. I suspect magazine and newspaper
Sign In or Register to comment.