Embedded Metadata: missing metadata options

Hi Folks.

I am working with a client to improve (amongst other things) citations to their site. At the moment, I am attempting to add enough embedded metadata to their pages to support Zotero citations (amongst others), using the instructions at https://www.zotero.org/support/dev/exposing_metadata .

There are two areas I am struggling with: the limitations of the "highwire" citation tags, and their interpretation by Zotero: I'd appreciate any advice that folk here might offer.

The highwire format is, of course, geared towards journals. Much of my client's content is based on books, and those books themselves are collections of primary documents (memos, telegrams, letters, minutes, etc). There are some fields that seem to be missing, but which are allowed by Zotero:
- [citation] Item type (see below)
- Series Number
- Place [of publication]

And others which Zotero itself is missing:
- Webpage created date (as recommended at https://www.scribbr.com/chicago-style/website-citations/)
- Webpage last updated date (ibid)
- e-locations (i.e. non page-based location references)

How should we be presenting these for Zotero?

Secondly, some of the citation tags which are supposed to be supported are showing up under the 'Extra' field when saved to zotero from the web browser plug-in:
- ISBN
- Pages
- Publisher
- Volume
And editors are being incorrectly captured as authors.

I suspect that, in part, this happens because there is no way to specify the citation type, and citations which should be interpreted as e.g. book sections are being captured as web pages. How could we correct this?

Finally, here's an example development page for illustrative purposes only: http://clementine.evolvedbinary.com:8080/exist/apps/hsg-shell/historicaldocuments/frus1969-76v19p1/d75
  • (gentle nudge)

    We may also be able to offer some dev time to resolve these, if someone could point us in the right direction.
  • You've seen https://www.zotero.org/support/dev/exposing_metadata I assume?

    For more technical details, Zotero's metadata import is performed in these two scripts, with the former calling the latter:
    https://github.com/zotero/translators/blob/master/Embedded Metadata.js
    https://github.com/zotero/translators/blob/master/RDF.js

    For your questions:
    1. Zotero can't make up metadata tags that don't exist, so where specific fields are missing, there's not much to do unless you can find some (semi-)authoritative description of them

    2. For webpages, Zotero just captures one publication date; it doesn't distinguish that from the last update date (yes, that does occasionally come up in citation styles, but it's generally not particularly useful). I'd use last update as the date. E-locators should typically just be pages, but are also rarely included in citation metadata (instead, they're part of a specific citation, i.e. would be added every time an item is cited, not in its general metadata)

    3. If fields like IBSN appear in Extra, they're not part of the item type selected (e.g., you have a webpage with an ISBN). They'll import properly with a correct item type.
  • Thanks for the reply, Adam

    Indeed I had seen the page (and linked to it in the question)!

    I'll take a look at those scripts; very helpful to get a starting point, thanks.

    1. Part of the problem seems to be that there is no "authoritative description" of the google/highwire fields - particularly for non-journal scholarly content!

    2. I think we can live with only one date; non page locations are more important for this sort of content (which will normally be a specific citation).

    3. So how do we publish an item type for Zotero to recognise?

    It feels like embedded metadata in this form might not be the best or most rewarding way to populate Zotero metadata: is there a better way?
  • 1. Right, that's why I say "semi" -- if it's used by a major repository or otherwise widely used, we'll implement it.

    2. As I said -- you can put them in the pages field if you want to, but maybe familiarize yourself a bit more with how citing in Zotero works, first: You typically insert a citation to a *work* and then, individually for each citation, specify a locator (i.e. page, e-locator, column, etc.). Those locators are not saved alongside the item metadata but rather input individually by the user every time an item is cited.

    3. Depends on the item type -- that's why I linked to the code. E.g. for books, setting dc.type to "book" works well.

    The most reliable ways to get Zotero to import exactly what you want are unAPI or writing a custom translator. See the linked page on exposing metadata.
  • Thanks,

    2. I do understand what you mean about how citing in Zotero works, but in this case the work is embedded in another, larger set of volumes. The location is important in identifying the work itself, not just the location of a given quote.

    3. I did try to get DC.type working on the page (see the link above), but sadly zotero still isn't pulling in the right citation type. I will look at unAPI or possibly MARC/MODS next. Again, thanks for your help.
  • edited May 18, 2022
    3. It's the og:type website that's throwing Zotero -- I'd argue that's a bug in the priority of item types Zotero uses (DC.type is more meaningful than og:type) but if you're able to remove or change the og:type it'll recognize this as a book (you can also get it to recognize it as a book section, btw. if that's more accurate)
  • Brilliant, that's fixed it. The og:type is there for social media previews on e.g. twitter; we probably ought to get it to be a bit more specific anyway.

    Thanks again.
  • PS: an og:type of document seems to work, too.
Sign In or Register to comment.