Zotero friendly metadata on custom site

Hi all,

I'm adding metadata to a custom-built site and have hit a few questions that hopefully someone here may be able to answer.

I've been experimenting with Dublin Core meta tags and with Coins, and am unsure yet which I'll use (or perhaps both).

Dublin Core with DCTERMS.bibliographicCitation

DC recommends including a machine readable citation in a DCTERMS.bibliographicCitation metatag (i.e. a ContentObject, the same as Coins format). However, when I try this Zotero ignores this and only reads the DC tags.

Will Zotero only read context objects in spans?

DC format in Coins

I noticed on the Zotero blog that the Coins tags use a 'DC' coins format, rather than 'Book' or 'Journal'. The Coins site only mentions these two, but I tracked this down to a proposed standard. Is this safe to use more broadly - i.e. supported elsewhere?

DC type list

Finally, I've been looking for a good controlled vocabulary for DC types. The Coins on the Zotero blog uses 'blogPost', but this doesn't seem to be from any particular vocabulary.

I tested different types by trying things from the "Item Type" drop-down list in Zotero. Is this a standard list, or Zotero specific? Or is there a good controlled vocabulary with a similar list of types that I should use? (The DCMI Types list is quite limited).

Thanks for reading. I'd appreciate any suggestions anyone can offer.
  • First of all, while it doesn't directly answer your question, you've seen https://www.zotero.org/support/dev/exposing_metadata ?

    Then
    Will Zotero only read context objects in spans?
    Yes. Do you have a link to that DC prescription for bibliographicCitation? Do they describe the format for those citations? We could certainly look at implementing that if it's going to be sufficiently standardized.

    As for your other questions about COinS -- the standard is woefully weak/underspecified the moment you get beyond books, so IIRC we intepreted it liberally to allow export of all Zotero types. Specifically, where no reasonable DC type exists, we use Zotero item types on export (and then also on import). Hope that makes sense.
  • Thanks for getting back to me. I have had a look at that page, and it seemed like DC or Coins were the easiest options.

    Regarding context objects in DC entries, see this page: http://dublincore.org/documents/dc-citation-guidelines/ (there are no anchors, but scroll down to "2.3 Machine Parsable Citations" - Example 5 shows the tags). It's the ContextObject format, but in a DC tag instead of a span.

    I see the COinS spec is a bit limited (and seems to have been left to age), however it still seems the easiest option compared to learning MODS.

    Thanks for the info regarding DC types. Their vocabulary is very limited - that's why I'd hoped you had found a better alternative standard :)

    It's the first time I've looked into this, and I'm surprised that the standards are so disparate and incomplete. Poking around various sites, it seems Zotero uses custom parsers for a lot of big sites, but I'd hoped there would be a reasonably easy open standard out there.

    Would you have any recommendations as to what to use? Should I be going for something more complete like MODS?
  • Yes, indeed, the standards are awful. I think that's partly because software working within the browser is really quite new. Remember, Endnote used to work exclusively in it's own search interface, driven by Z39.50.
    People are working on RDFa for bibliographic data, that might go somewhere (no Zotero support yet, though).

    As for the best standard to pick -- it somewhat depends on what type of bibliographic items you have. For mostly typical bibliographic stuff, google highwire tags are great and, within that scope, very expressive and accurate (and also give you good google scholar results for your site).

    If you're going to have more unusual item types, unAPI with MODS (or even RIS) give by far the best import results, but afaik only Zotero uses unAPI. COinS et al. are also read by other software, including the Mendeley bookmarklet.
  • Thanks adamsmith.

    I saw the highwire format, and it does seem the closest to what I was expecting to find - i.e. an easily implemented set of meta tags that cover most cases. But, it seems like it's not a very open or well-documented format.

    To be honest, it's as much an exercise in curiosity than practicality adding these - the site is a small archive, and many items don't fit obvious bibliographic types too well anyway. For example, we have a number of full issues of old publications (papers/magazines), which won't map to a Zotero type anyway.

    (As an aside, COinS has a format for journal issues, but Zotero only accepts individual articles among it's types, so it quietly fails on these. Would there be any call for a Zotero type like this? The book type won't cover volume and issue numbers).

    I'd rather have something as generic as possible, so even though unAPI would give the most scope, if it's not used elsewhere I'm not sure it's worth learning something as extensive as MODS.

    I'll try some compromise of DC, with COinS where they're useful, I think.

    Regarding that DC recommendation for ContextObjects above, would it be worth opening a ticket on the embedded metadata extractor to point this out? Or maybe it's not actually used in the wild anyway.
  • I've never seen bibliographicCitation in the free-form or tex.KEV format. I doubt it's worth the effort.
  • Thanks, aurimas. I figured if you hadn't come across it here, it's not a standard that's been taken up.
  • (we'll introduce a periodical item type -- not sure about individual journal issues; apart from special issues I wouldn't think those are ever cited).
  • True, I suppose there are rarely occasions when one would cite the whole issue rather than an article. What do you mean by a periodical type? A reference to the journal as a whole, rather than a specific issue?
  • correct, yes. That has been requested rather frequently.
  • We have an SFX customer that has been asking for way to store journal level data in Zotero from the SFX link resolver (SFX service menu has coins - OpenURL KEV format). How can we know when this periodical item type becomes available? The customer would be very interested in this.

    Lieve Rottiers - ExLibris Ltd. - SFX analyst
Sign In or Register to comment.