PubMed (NCBI): Doubleclick to open link that was saved from PubMed doesn't open in PubMed

edited November 27, 2018
This is disorienting. Opening link that was saved from PubMed (NCBI) doesn't open in PubMed, but opens in the journal that published the article. This is very disorienting since the user used PubMed to save the link, and can always reach the journal from the PubMed record.

How do we fix this so that references saved from PubMed opens in PubMed and not the original journal that published the article?
  • Really depends on which link you're talking about.

    Zotero does attach a link to PubMed that will go to pubmed. If you expand the item in Zotero and click on the attachment labeled "Pubmed Link", it'll open in PubMed.

    Zotero also tries to access an Open Access PDF, which would typicallly be from pubmed central in this case, so you might be seeing that.

    If you're actually seeing a journal webpage opened, though, you'd have to click on the DOI label in Zotero and I wouldn't think that's what you're doing since it'd have to be pretty intentional.
  • When you click on the item (not the Pubmed entry or any attached pdfs), it does take you to the webpage specified by the DOI. At least that is the current behavior in my hands with Zotero 5.0.58.

    To go to the Pubmed webpage, you have to click on the child Pubmed entry.

    I also think this makes a ton of sense.

    The older Pubmed connector used to save the Pubmed URL as the webpage (I think), which did not make much sense, and that has been corrected a long time ago.
  • Dear adamsmith and enolzkan,

    Thank you both. My Zotero 5.0.58 was not creating a child entry with PubMed link, and after I reinstalled the browser connector, it did make the child link to PubMed. I think it would be less confusing if the main link (not child/children) pointed to the exact same location where the entry was made originally. This is how Zotero used to be (doi link was the child), and it was less confusing for the users (in my humble opinion).

    In today's world, we are often oversimplifying/overthinking our users. Sometimes less automation is more usability.

  • I think it would be less confusing if the main link (not child/children) pointed to the exact same location where the entry was made originally.
    That would be incorrect. These fields aren't arbitrary. The URL field gets cited, and the URL to cite is the canonical version available from the journal, which is also where the DOI takes you. Double-clicking the item resolves the DOI if there's nothing in the URL field, which is likely why you're ending up there.

    The attached PubMed link here is equivalent to any other external database, such as Amazon for books. You don't cite the Amazon URL, so it doesn't belong put in the URL field and is added as an attached link instead.
  • Dear dstillman,

    Thank you for quick rely, FYI, after checking carefully off an on for days, I just realized that FireFox Tabo addon (https://addons.mozilla.org/en-US/firefox/addon/tabo/) seems to prevent creation of PubMed child record. I have over 20 addons so it wasn't easy to figure out ....

    As you know, such conflicts have happened before (e.g. https://forums.zotero.org/discussion/66924/firefox-54-plug-in-conflict-with-zotero-5, or https://forums.zotero.org/discussion/68684/zotero-connector-conflicts-with-lastpass-add-on). I have also seen it before when using ublock origin, which resolved when I disabled ublock and it was not a big deal for me.

    I really hope the Zotero can do something to prevent such problems, or at minimum keep an online record (e.g. json file or database) where Zotero can check and warn the users regarding conflicting addons they have in their browsers.

    Thank you for your awesome work :)
  • edited November 27, 2018
    Dear dstillman,

    Just in my humble way of thinking:

    PubMed = NLM, NIH, Goverment authority focused on building reference library.
    Amazon = An online store, albeit big one.
    Google Books = Private reference source (perhaps somewhere between Amazon and PubMed)

    I just think there is an important difference with the way scholars, students, and learning folks think about them. So it is not so simple, specially since some journals may completely close/alter the link or go out of publication/business, but PubMed is likely to survive longer. If the original/parent link points to the location a user used to grab the reference, he/she will remember much easier when/why/how he/she grabbed the reference for their Zotero database.

    Maybe, I hope there will soon be a way to give user's a choice in this regard by a setting/preference.
  • by Google, I meant "Google Books"
  • Dear dstillman,

    It just happened again, missing child records, even with Tabo disabled. This is likely a bug with the new FireFox, and seems to work when enabling/disabling some addons, but doesn't stay for some reason, specially when changing from local to group libraries. I will try and see how I can reproduce it reliably, and I will post once I find it ....
  • Dear dstillman,

    I confirmed and reproduced it reliably several times. It is most likely a bug, child record is not made when using Group Libraries (tested only with PubMed).

    I tested it using clean new install of Windows 10 and 7 in virtualbox environment, and clean/new install of FireFox 60.3.0esr (no addons except Zotero) and Zotero 5.0.58. When using Group Library, refresh a PubMed page and try to save it to Zotero. You will see child record is missing. When you change back to own library (default name "My Library"), and try to save without refreshing the PubMed page, the child record is still missing. Once you refresh the PubMed page (while using "My Library"), then the child record is made again.

    Please feel free to move post and/or create bug report as needed. Thank you for helping to fix above problem asap, it is very important for us.

    Additionally, if we close Zotero standalone, and enable saving to online library, it saves to My Library and not the intended Group Library (perhaps another bug or needed feature)
  • It is most likely a bug, child record is not made when using Group Libraries (tested only with PubMed).
    Sorry, looks like I never followed up on this, but this is fixed in 5.0.60. Web links erroneously weren't saved to group libraries that didn't allow file attachments.
  • Totally agree that the default behaviour for clicking on an item in a library should be to open where it came from. This is the obvious discoverability behaviour. I specifically save to zotero from the pubmed URL and always prefer to open that -it's tons, tons, tons more useful to me than a publisher's site -it's also faster, has no surprises, and is consistent.

    I think maybe there's some conceptual conflation in using zotero as a research tool vs as a store for bibliographic data. The logic of the store and the interface should be separate. I would guess 99% of the time zotero is used as a research tool and so should probably have defaults optimized for that.

    Having said that, I'd be really happy if I could at least configure the behavior I want.
  • @zotmik I don't want to reiterate Dan's point above why your suggestion is not expected behavior (the argument with why you wouldn't cite an Amazon page for a book).

    But at least consider this: Many of us use Pubmed as our bibliographic source, since it includes high quality data. But we still go to the journal's page to read the actual paper. Once the bibliographic information is saved, you nearly never need Pubmed, but still need the article.

    Also, you can still go to Pubmed's page quickly: Click on the triangle next to your saved item -> Double-click the "Pubmed entry". That's just one extra single-click compared to your proposal.
  • @enozkan actually my point is that I don't use the zotero interface for citing -there are wonderful connectors for that. I expect the connectors to take care of using the right information for citing as that is their job. The zotero interface behaviour is not really related to this, the semantic meaning of the data, or to how the data is stored. Dan seems to be arguing that the interface has to follow the semantic meaning but these are all different things. When I click save to zotero from the browser and then I click on that item to open in the browser, it is unexpected that it should go to another site. I have to google to figure out what happened and that is exactly what discoverable design isn't. I don't think anyone will ever have to google to figure anything out if clicking on the item opens the page they saved.

    Anyway, I see we prefer different behavior. I really have no need to go to the publishers site from zotero. If I want the paper, I go to my associated pdf -which I love about zotero. It may sound like winging but to me 50 extra clicks a day is fairly big. Would you be happy if the behaviour switched back to what it used to do so that you had to make the extra clicks?

    As I mentioned though, I'd be super happy if I could toggle this behavior.
  • By the way, there was a time the Pubmed translator used to save the URL for Pubmed (exactly as you would like to see it happen).

    After it was "corrected", I realized that this made a lot more sense for nearly everyone. There is little reason anybody would want to visit a catalog's page twice (Pubmed is not special in this regard), and yet that function is still conserved by one extra click.

    If you would like to see the old discussion, check it out here: https://forums.zotero.org/discussion/25635/pubmed-import-partially-broken/p1
  • Dan seems to be arguing that the interface has to follow the semantic meaning but these are all different things.
    It's not just about what's cited — what's cited is a proxy for the conceptual home for the item online. You seem to be thinking of Zotero as a bookmark manager, but that's not what it is. You're not saving a webpage — you're saving the abstract idea of the journal article described on that page, and you can do that from various places, including Add Item by Identifier, which doesn't come from a specific URL. You say that when you want the paper, you go to the associated PDF, but that's exactly what Zotero does when there's a PDF attached to the item. When there's not, it takes you to the next best thing, which is the canonical home specified by the DOI or URL. (And the PDF itself doesn't necessarily come from the page you saved from, so this behavior is also consistent with that.)

    We won't be changing this or making this configurable, but you or somebody else could create a plugin to open the first-created link when one is available.

    We'll also be adding a PMID field to the right-hand pane in an upcoming version, and clicking that label should probably take you to the PubMed page. (At that point, I think we could stop attaching the link altogether.)
Sign In or Register to comment.