After zotero?

The charity I work for is considering using Zotero as a knowledge management tool, to build a shared library. There are other software tools we are considering, but Zotero stand alone has several advantages. One of the contingencies we are considering is what would happen if we build a library and then for whatever reason decide we should move to a different package. We want to make sure that we could take our library, collections, tags, and pdfs to the new package.

Does anyone have any experience of this? I guess if you're on here it's because you love zotero though. Any thoughts about this would be much appreciated!

Thanks,
Sarah
  • The ease of getting your data out is one of the core strength of Zotero. It exports in all standard reference formats like RIS and BibTex which can be imported by all reference managers and also supports some more expressive standards like Bibliontology RDF and MODS.
    You will typically lose collections on export--> import, though, unless you export them one-by-one.

    Zotero's database is easily readable - Mendeley, e.g., makes use of that and allows you to (one-way) sync your Zotero database into Mendeley (that includes collections).

    Finally, Zotero has a well specified API that allows any third party developer to sync a Zotero library. Currently that's only used for mobile apps for Zotero, but it could also be used by any other tool to import Zotero data.
  • Thank you Adam. Yes it's great that Zotero can export metadata. We're particularly interested in its strengths at exporting collections and tags.

    Can you give me further information about that? We are looking to set up many collections within our library so exporting them one-by-one would be too laborious. Can the API be used by other interfaces to import that information? How else could it be done?
  • The export part will not be a problem. Importing the data may be a problem if the reference management software that you migrate to will not support importing tags, or will not support collection hierarchies, or something else.

    If you decide to move to some other references management software, people here will definitely help you in getting your data out of Zotero.
  • You will typically lose collections on export--> import, though, unless you export them one-by-one.
    Is there really no way in either RIS or BibTeX to encode that an item belongs to one or more hierarchical collections? Since most modern reference managers now support the concept of collections, is it not possible to extend these formats to handle this? Is Thomson-Reuters the caretaker of the RIS format?
  • RIS has no concept of collections, but we have tags (keywords) that could possibly be hacked into supporting collections (something like KW - collection:col1/subcol1/subcol2). And yes, TR is the maintainer of the RIS "specification"

    BibTeX is more or less in the same boat, but we (mostly just @friflaj) did just add support for importing collections from JabRef, which uses a special item type that would otherwise be ignored by all other managers to specify the collection hierarchy and assign items to it. See http://github.com/zotero/translators/pull/603
  • edited September 13, 2013
    There are five ways to export collections from Zotero

    1. Directly reading the zotero.sqlite
    2. Exporting to Zotero RDF
    3. Exporting to BibTex (once the pull request referenced by aurimas goes through, which it will shortly)
    4. Via the server api: http://www.zotero.org/support/dev/server_api/v2/read_requests
    5. Via the local javascript API: http://www.zotero.org/support/dev/client_coding/javascript_api

    As far as I know, no major competitor offers anything beyond 1. and 4. and not even all do that (e.g. I don't think there is any legal way to do this for RefWorks or EndNote)

    Of these
    1. is implemented in Mendeley
    3. is designed to work with JabRef

    So these are currently the only programs I'm aware of that you can easily import Zotero items with collections into, but any of those five solutions could theoretically be implemented for a target software (though that would have to be done by either you or them), it just depends on the capabilities of the software you're importing into.

    Edit: See Aurimas below on JabRef - we may still implement this, but probably not anytime very soon.
  • One slight correction. The BibTeX/JabRef support is currently only for importing into Zotero, not exporting.
  • I figured it couldn't hurt asking on the EndNote forums if there are any plans to advance the RIS standard and support collections. See http://community.thomsonreuters.com/t5/EndNote-Product-Suggestions/Update-RIS-format-to-support-groups/td-p/46987
  • Quick note: Paperpile uses method #4 (server API) to import collections and attachments from Zotero. It's working well for us, but the effort needed to implement a full library import via HTTP APIs is frankly too high.

    It's pretty damning for the community that there's still no way to reliably share the most common denominator of data types (library entries, attachments, tags & folder hierarchy) between different programs using a simple file format that users can work with.

    We'd be very supportive of import / export of hierarchical collections if a suitable standard were to arise. I personally don't like the JabRef approach (it feels like a hack trying to put such detailed info into comment fields, with a very convoluted syntax as a result), but if that gains wider adoption it's something we could support.

    More interesting would be to have a .zip file format which contains a Bibex or RIS file, all attachments, and a file describing the folder / tag hierarhcies. One could use something like OPML to describe the hierarchy, with contained items referenced by their library IDs.

    Zotero's exporting of attachments + RIS files is close to this — but misses entirely the folder hierarchy.

    Paperpile has only just started, and we plan to support more import / export options in the future. It would be great to move toward a more sane interoperable community standard.
  • @gjuggler: please don't forget legal support. It's a hard case, without clear international standards yet, but everybody cites it at some point.
  • FWIW, I think if the community were to adopt a unified format, I would steer away from RIS and BibTeX, since these formats would have to be modified/extended anyway. IMO, an XML based format, like Bibliontology RDF, would make the most sense in terms of flexibility and structure. It still needs some work, but it's a pretty good place to start.
  • @fbennet: thanks for the reminder! We've had legal support on our minds for a long time, but we forced ourselves to focus on what we know best (biomedicine) at the start. The publisher support and UI for a streamlined legal setup would be significantly different than what scientists require. That being said, we've followed along the legal developments closely and find lots of potential in that arena. (It's crazy when you talk to working lawyers, how much of it is done manually with reference to that famous colored book.) On a side note: we'd love to start running the citeproc-js tests locally to be able to verify bugs or add new test cases. Is there an up-to-date guide on how to run the tests?

    @aurimas, RDF may make the most sense in terms of flexibility and structure, but the breadth of existing support for RIS or BibTeX can't be discounted. If a complicated an relatively unsupported file format were adopted, this would mean Yet Another Format for every reference manager to implement support for (i.e., not gonna happen). On the other hand, agreeing to extend an existing format in a sensible way would require minimal changes all around, which I think greatly improves the chances of it actually happening. JabRef's group support is an example of this practical approach. I personally don't love the syntax, but it *is* better than nothing.
  • Whoops — note, the above post by "pp.staging.test" was written by me (gjuggler). I forgot to log out of my testing account...
Sign In or Register to comment.