[citeproc bug] wrong disambiguation default

According to the specs, givenname-disambiguation-rule should default to "by-cite".
In the current citeproc version in Zotero 4.0.4 the default seems to be "all-names" and, even worse, it seems to ignore the name formatting settings:

See here for a test set, use e.g. with Elsevier Harvard:
https://gist.github.com/adam3smith/5391997

You get:
(Hall and Kathleen A Thelen, 2009)
(Steinmo et al., 1992)
which shouldn't be possible both because disambiguation isn't necessary _and_ because if it's going to disambiguate it should do so with initials.

Original thread:
https://forums.zotero.org/discussion/28868/elsevier-citation-without-first-names/#Item_9
«1
  • There is only the one bug. It should be defaulting to by-cite, but it's using all-names instead. I'll check into it.

    The disambiguation is not using initials because the sample data has "Kathleen A" and "Kathleen Ann", so it renders the full name to distinguish the two (despite only one appearing in a cite -- that's the all-names rule at work).

    If the form of the name is aligned, you get (Hall and Thelen, 2009), with both the by-cite and the all-names rules.
  • ah yes, that makes sense on the initials vs. full name.
    I realize of course that the names should be aligned in correct data, it was just the first example that I could come up quickly.
  • It's fixed now. The tests were all passing, so I missed the fault (for a long time -- sorry for the frustration over this). It was initialised badly, which only seems to have caused problems when running in the client. You can run the new code using the patch plugin.
  • I realize of course that the names should be aligned in correct data, it was just the first example that I could come up quickly.
    The sample was well-chosen -- the misalignment was the key feature that exposed the bug.
  • Thanks! Still doesn't look right for me in the test pane running the 1.0.86 version of the gadget, though.
  • Uh-oh. I'm off to work now, I'll have another go a bit later.
  • The tests I built this morning must have been badly constructed. I've found the actual source of the bug now. Unfortunately it's going to be down the rabbit hole for me again on this one: fixing the problem has caused things to come unglued elsewhere.

    When it's fixed and stable I'll post again, or write privately. But it may take several days to sort this out.
  • I have good code for this now. There will be some impact on the processor test fixtures, some of which assumed the former all-names default.

    Changes to test results means changes in user documents, so once I get things tidied up, I'll post a set of links for inspection before pushing out the new processor version.
  • The fix is now available. To try it out, install this plugin. (All it does is to swap the revised citation processor for the one contained in Zotero. You can uninstall it when the next Zotero release comes out.)
  • edited April 17, 2013
  • The three last links are dead - they're missing a humans/ before the filename (I looked at them, though).

    This all looks great to me. I'm confused about one of the tests:

    https://bitbucket.org/bdarcus/citeproc-test/src/tip/processor-tests/humans/disambiguate_AndreaEg1b.txt?at=default

    Smith, Brown & J. Jones (1980); Smith, Brown & J. Jones (1980)
    Smith, Brown & J. Jones (1980); Smith, Benson, et al. (1980)
    Smith, Brown & J. Jones (1980); Smith, Brown & A. Jones (1980)

    I understand the third test, but neither why J. Jones is listed in the second (wouldn't Smith, Brown, et al. and Smith, Benson, et al. be sufficient?) nor why he's listed in both items in the first row - that doesn't seem to add anything - shouldn't those just be Smith et al. (1980)?

    I'm probably missing something simple, but if you don't mind clearing this up for me?
  • oh - and tested the fix - works great now.
  • Oops, thanks. Fixed those links.

    The cites are rendered in pairs, but as if in a single document. The last cite is with A. Jones, so the others need to be expanded to distinguish it.
  • ah ok, makes sense - thanks!
  • Will that fix work on a mac? if so, how do I use it?

    And if not, is there anything else I can use to avoid changing all the citations by hand? I have about 3 days to figure this out before I have to do that.
  • do you use Zotero for Firefox or Standalone?
  • Standalone.
  • I don't think the plugin works on Standalone, though you can try:
    Download:
    http://gsl-nagoya-u.net/http/pub/zotero-processor.xpi
    (right-click --> save link as...)
    Then install the downloaded zotero-processor.xpi in Standalone via tools --> add-ons --> click on the screwdriver/wrench icon --> install Add-on from File

    Restart Standalone and, to be sure, switch to a different citation style and back in your Word document.

    There's also a decent chance that Zotero 4.0.6 will come out tonight, though Dan would have to say.

    If all of that fails it's possible to manually adjust the style - wouldn't be very hard either, but let's talk about that tomorrow if the first two options fail.
  • edited April 21, 2013
    First, thank so much for your speedy replies!

    I tried everything you suggested. The plugin, as predicted, didn't really work.

    I think I wasn't actually clear on the problem, and I'm not sure how to phrase it.

    Here's an example of a citation I'm having trouble with. it's supposed to be in APA format.

    (Benigas & Bourgeois, 2011; M S Bourgeois, Burgio, Schulz, Beach, & Palmer, 1997; M S Bourgeois, Schulz, & Burgio, 1996; M S Bourgeois, 1993; M. Bourgeois, 2006; Michelle Bourgeois, Fried-Oken, & Rowland, 2009; Michelle Bourgeois & Mason, 1996; Michelle S. Bourgeois, Dijkstra, Burgio, & Allen, 2004; Michelle S. Bourgeois, 1991; Michelle S Bourgeois et al., 2003; Michelle Bourgeois, 1990, 1992; Chang & Bourgeois, 2012; Dijkstra, Bourgeois, Burgio, & Allen, 2002; Hoerster, Hickey, & Bourgeois, 2001)

    Initials with and without punctuation keep showing up, along with entire first names, which have absolutely no business being there at all. This has something to do, I guess, with having used the Zotero Chrome plugin to import the references in the first place.

    Is the only way to have each citation look like "(Bourgeois, 2007)" or bibliography entries to be "Bourgeois, M." rather than "Bourgeois, Michelle S." to go in and change every one? Is there any way to standardize a given author's name?

    (I hope that asks my question more clearly. I look forward to the new update and really enjoy using this program!)
  • Well, generally you may not get correct citations/bibliography with imperfect data - i.e. Zotero will treat the same author with differently entered first names as different authors so I'd strongly encourage you to fix that in your database.

    That said, the Elsevier Harvard style should only ever show initials where they're strictly necessary to disambiguate two citations that would otherwise be identical, so in the fixed version, none of your examples would have initials or first names.
  • Thanks! It's good to know I can just take care of it instead of forever wondering if there's another way.

    I'll look into the Elsevier Harvard style as well.
  • edited April 21, 2013
    Oh sorry I don't think you mentioned APA in your post originally, I assumed you were using Elsevier Harvard already.

    The 4.0.6 update will _not_ fix this for you (nor will any future update). APA style does require to disambiguate primary authors for in text citations, so this is working exactly as desired. If you need APA style there is no alternative to cleaning your database.
  • @skl42,

    As adamsmith says, this is an issue with the way names are recorded in your database. If names for the same person are recorded inconsistently, you will get the kind of output (with extra initials, etc) shown in your sample.

    The problem comes from inconsistency in the original data. It doesn't matter what tool you use to fetch items; if publishers and aggregators provide a name in different forms, the difference will show up in your Zotero database, and needs to be fixed by hand. If you curate the records from time to time, it can save time in the long run.
  • I have the same problem with Emerald format. the source of issue comes from the way zotero collects the name online. and I am unable to fix it! can I send you my whole document to see all the situations zotero messed stuff?
  • no need for the document, no. What version of Zotero?
  • 4.0.6
    it seems it messes up middle names and multiple names a lot and the source is in the library and not the reporting part. is there anyway to fix it.
  • For a start upgrade to 4.0.8, though I don't think that will make a difference.
    You'd have to provide some examples for us to see, but as we say above, you do need to manually adjust/unify author names for entirely correct citation behavior and there is really no way around that.
  • lets say this link

    http://link.springer.com/chapter/10.1007%2F978-0-85729-189-9_13?LI=true

    when you capture it with zotero... and then insert it into the text with harvard format it brings all first, middle and last name of author rather just last name..... I guess its might be coming from how zotero can not identify last name over other names in spie of the comma in the library. or the style reporting plug-in does not know it.
    in both case you can see if affect all my work with above 80 citations.

    Is there any document pointing out cases that one needs to be careful when using zotero..its pretty painful to discover it by mistakes that reviewers find.
  • sorry, still confused. When I import that item and then cite it with the Emerald Journal style I get
    (Hockley et al., 2011)
    as an in text citation and

    Hockley, C.J., Smith, J.C. and Lacey, L.J. (2011), “Contracting for Availability and Capability in the Defence Environment”, in Ng, I., Parry, G., Wild, P., McFarlane, D. and Tasker, P. (Eds.),Complex Engineering Service Systems, Decision Engineering, Springer London, pp. 237–256.

    which looks good to me.
    That said, there is no document, but as a general rule, Zotero simpifies referencing, it doesn't remove you from the task entirely. You do need to check that data imported in Zotero is correct, you should verify that citations created by Zotero conform to those required by the journal, and, as we say above, you need to check that your authors are imported in a unified form. This is true for all reference management software.
  • contromir: The name of a given author needs to be set consistently for each citation in your document. If that's done, the software will format the names correctly. As adamsmith says, this isn't a bug in Zotero or the citation processor: you just need to have clean data to get clean output.
Sign In or Register to comment.