Available for beta testing: Create parent items by identifier

edited November 21, 2020
In the latest Zotero beta, we've extended Create Parent Item to prompt for an identifier (DOI, ISBN, PMID, arXiv ID) to use to retrieve metadata for the new parent item, making it easier to create a parent item when metadata can't be retrieved automatically for a PDF. If the publication doesn't have an identifier, you can still create an empty item and enter the data manually.

This feature will be included in Zotero 5.0.94.

(We're also working on metadata updating for existing items, but that will be a separate feature.)
  • I gave this a spin with DOIs and it's working very nicely!
    Have you considered allowing URLs there & using translator infrastructure to import metadata? That'd seem to likely be technically possible basically doing locally what zbib does?

    FWIW, I don't know if I buy that having separate processes for "create parent from PID" and "create new item, paste in PID, update metadata" makes sense. I think just allowing update metadata is sufficient as it allows you to do the same and is simpler, but I'm not great at predicting what UX elements work, so...
  • Have you considered allowing URLs there & using translator infrastructure to import metadata?
    We have, and it's a possibility. We don't support URLs in Add Item by Identifier largely because gated files wouldn't be available, which isn't an issue here. But gated databases and databases without permalinks still wouldn't work, so I'd be a little worried that it would make things more error-prone and confusing with limited benefit. If there's a PDF and useful extractable data, there's a good chance there's also an identifier on the page. But I'm open to allowing this.
    FWIW, I don't know if I buy that having separate processes for "create parent from PID" and "create new item, paste in PID, update metadata" makes sense
    I'm not totally sure what you mean for the second option here. Do you mean just having "Create Parent Item" work like it used to, creating an empty item, and relying on pasting and updating, or do you mean creating a parent item automatically?

    If you mean the former, I think a dedicated window like this, where "Create Parent Item" privileges creating by identifier, is a clear usability win over creating an empty item and hoping that people think to paste in an identifier in the correct field (which currently won't even exist much of the time) and trigger metadata updating. Metadata updating is really intended for a different use case, where you have existing metadata filled in and want to update it, and the implementation we're working on offers granular control over field changes, which isn't necessary here. We can optimize pasting an identifier into an empty item — automatically retrieving metadata and skipping the diff dialog — but it's still not really a replacement for a simple dialog that asks you to paste in any identifier.

    If you mean automatically creating a parent item for PDFs that can't be recognized, that could make sense. It would obviously avoid a very common source of confusion. My main concern is that a significant number of people who don't care about citations do use Zotero with standalone PDFs, and I'd be a bit nervous about enforcing the creation of parent items with just a title named after the PDF filename. It would also make it more awkward to add parent items via web or import for unrecognized PDFs with no identifiers (though allowing adding by URL would address much of the web case). But it might be worth it — there are advantages to always having a parent item, including the ability to take any number of child notes instead of relying on a single embedded attachment note (which we've always sort of regretted supporting). This would still have the problems I mention above about pasting an identifier into a specific field, so I think we would still want some sort of option to set metadata by any identifier, either in the context menu or even in the item pane for otherwise empty items (if we could do so in a way that wasn't too annoying for cases where an identifier wasn't an option).
  • Thanks for the detailed answer -- I agree that the hope would be that the scenario where the URL works but the PDF isn't recognized is rare. I'd expect the most likely scenario are working papers, especially those published without DOIs (i.e. not found by CrossRef) but on sites with good metadata. E.g. the PDF on
    https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1450343 was in my library from when it came out isn't recognized by retrieve metadata, but zbib finds great metadata for the URL. I think URL is worth including as an option.

    For the adding via identifier scenario, I was actually thinking of your first option. My thinking was that this is somewhat rare and the separate dialog seems unecessary for that, but as I say, I could be completely wrong.

    Automatic creation of parent item if no match is found -- I wasn't thinking about this and I see the same pros and cons you do. I think for the people just using Zotero as a PDF manager without interest in metadata, a pref would likely do, but the other concerns remain.
  • For the adding via identifier scenario, I was actually thinking of your first option. My thinking was that this is somewhat rare and the separate dialog seems unecessary for that, but as I say, I could be completely wrong.
    Oh, I don't think it's rare at all — we tell people all the time, and in the documentation, that if the PDF isn't recognized they can either save from the web or use Add Item by Identifier to create a parent item, and only if that fails enter metadata manually. So this is simply privileging that (particularly if we allow URLs) and deemphasizing manual entry. If there's no identifier, or you really want to enter metadata manually, an empty item is just one more click ("Manual Entry") in the dialog.

    So I'm actually pretty confident that this will avoid a lot of confusion and tedious manual entry. We just might be able to come up with something that also avoids the problem of people not knowing to use Create Parent Item in the first place.
  • There is already a pref to disable automatic metadata retrieval. So, for users just using Zotero as a PDF manager, presumably that is off and they would only ever invoke any of these options by clicking Retrieve Metadata. In that case, going directly to creating a parent on failure seems appropriate.
  • This is available now in Zotero 5.0.94. We'll work on further streamlining the process for unrecognized PDFs in a future version.
  • Hello,

    I am having a problem with "Create Parent Item"

    I have used this functionality when adding files from my computer for months, but today, when I click that option I am being prompted with an input screen to "Enter a DOI, ISBN, PMID, or arXiv ID to identity this file" to bypass this, I have to click the button "Manual Entry" for it to create the parent item.

    Before, this did not happen and I have not changed any settings recently. In my preferences, the box is not checked to "Automatically retrieve metadata for PDFs" and I have not changed my preferences recently. Before, I was able to add files and create parent without getting this prompt.

    I am using 5.0.94. In the release notes, there is a change noted for Create parent item: When right-clicking on an unrecognized PDF or other file and choosing Create Parent Item, it's now possible to specify an identifier (DOI, ISBN, PMID, arXiv ID) to use to create the parent item.

    How can this be turned off so that it just creates the parent item (without a prompt)?

    I do not want my default functionality to Create parent items by identifier.
    I want my default functionality to just Create parent item manually.

    Thanks for your help!
  • How can this be turned off so that it just creates the parent item (without a prompt)?
    Currently it can't, sorry. Not sure if Zotero would consider a preference there -- I'd honestly doubt it -- it seems like the type of thing where overcustomizability creates more problems than it solves (fairly rare use case, lots of additional troubleshooting & potential confusion).
  • But we're likely going to start automatically creating parent items when PDFs can't be recognized and providing another way to enter a parent identifier, so Create Parent Item won't be necessary.

    I'm curious what your use case is here, though, @CityPrincess.
  • My current work around is to use the prior version 5.0.92.

    @dstillman: Many of my files are already locally stored on my computer. I drag/drop them in batch to add them to Zotero; then I will typically select the batch of files that have no parent and click the option to Create Parent Items. Now I am able to finish adding the metadata I need such as tags or use some of my add-ons which typically only work from Parent Items and not on stand-alone files.

    With the new behavior, instead of being able to do batch operations, I would have to click on each file individually, and then click the create manually button from the pop-up window. That's a huge waste of time.

    It would be convenient if parent items were automatically created from adding files via the drag/drop; but at least in the prior versions the parents could be created in one batch.

    @adamsmith: The new functionality makes a lot of assumptions about how users have their files and data as well as forces a particular piece-meal workflow that is sub-optimal for some users. Since the new version is forcing users to select each file individually to create the parent item; an additional menu option could have been added to "Create Parent Item from Identifier" which would then show the prompt for the user to add the identifier. That could have been an addition that did not take away the functionally that existed up to v5.0.92 to create parent items manually (and in batch). I think having two menu items (1) create parent item manually - which can accommodate batch processing and (2) create parent item from identifier - which creates parent for a single item would not be so complicated and a better solution for all users.
  • The lack of batch functionality in the new implementation is a good point, that is indeed inconvenient. dstillman apparently agrees -- ticket is here: https://github.com/zotero/zotero/issues/1934 work on putting that (back) in already underway.

    (I don't think there's much of a chance for 2 context menu items, though -- that's not great UX)
  • We've restored the dialog-less "Create Parent Items" for multiple items in the latest Zotero beta.
  • @dstillman Thank you for re-establishing the ability to batch create parent items in the latest Zotero beta. Like @CityPrincess, I often drag a bunch of pdfs into Zotero (with filenames carefully curated so that I can find what I need), then I create parent items for all of them at once, and then I refine the metadata in Zotero (e.g., changing entry type from "Document" to "Book," etc.) as I have time/interest.
  • "Create Parent Items" for multiple items has been restored in Zotero 5.0.95, available now.
  • Thank you @dstillman! This was quite annoying because, like @bthomaslee I regularly import files that require manual entry, like scans. I'm happy to see how rapidly Zotero has responded to a drop in functionality whereas most big tech companies seem to ignore that sort of thing.
  • Thank you for restoring the batch-create functionality!
Sign In or Register to comment.