export annotations

adoramuste
while reading a pdf text, i compile lots of annotations and comments.
the problem is how to automatically export those compilations to other applications [say: scrivener, word, libreoffice].
thanks for any bits of advice.

  • dstillman
    If you're asking about the new PDF reader in the Zotero beta, you can add the annotations to notes, add additional notes or citations as you see fit, and then insert those notes into Word, LibreOffice, or Google Docs with active citations using the word processor plugin.

    You can also just copy and paste from notes to other applications, though the citations won't stay active.

    If there's some other workflow you have in mind, can you say more about what you wanted to do?
  • adoramuste
    yes i am working on the new PDF reader in zotero beta.
    thank you for your fast reply @dstillman.
    is it possible to insert my annotations/notes in zotero to word processor *without* copy-pasting it?
    [sorry with my silly question]
  • dstillman
    Yes, there's a new Add Note button — see the linked page.
  • adoramuste
    ok, thx!
  • sdspieg
    edited August 20, 2021
    But so Dan, if we want to just export the annotated text spans, how do we do that?
  • dstillman
    You'd create a note from the annotations in the Notes pane of the PDF reader and then either copy/paste or use Add Note in one of the supported word processors.

    Plugins such as mdnotes can also export notes, though I'm not sure if that one has been updated for the Zotero beta.

    We plan to offer additional direct export options for notes in the future.
  • sdspieg
    We now have a few 100s of notes across a library... So doing it manually is a lot of work. I'll look into mdnotes though. Thanks Dan!
  • dstillman
    Could you say more about the workflow you have in mind?
  • dstillman
    edited August 20, 2021
    You can also just do a search (or saved search) that matches notes, do a Select All, and then drag them to any text field to copy the text.
  • sdspieg
    I'll try that - thanks.
    As to our workflow - we are now using our Zotero libraries to hand-pick 'golden sentences' from a corpus (/Zotero library). So members of our team go in a library, select articles that, based on their titles, look particularly promising for a given research topic. We then open the pdf (now in the new Zotero tool) and highlight those precisely relevant sentences. We typically want 100-500 of these golden sentences. These can then be used to train models, using various NLP tools (in our case Prodigy/spaCy - 'active learning') that help us in discovering similar sentences in our entire corpus - and that corpus is often many thousands of pubs (/items in Zotero).
    Academic documents are not easy (yet) for most NLP. So golden sentences are a 'short-cut' that can help 'boot-strap' these semi-automated binary (relevant/not-relevant) classification models. In summary our workflow is:

    * keyword search query formulation (given the INSANELY poor search query syntaxes of most academic full-text publication databases, that takes a HUGE amount of experimentation, validation, documentation, etc.) ->

    * using that query (often containing 100 of terms with all sorts of search operators) to download a 'relevant' corpus that has both precision AND (much neglected!) recall ->

    * getting all of that somehow into Zotero (because we ultimately use it for footnotes/bibliographies - I have asked many times over the years to build more 'NLP' functionality into Zotero, but I understand that's not your core business as you see it) ->

    * finding and highlighting 'golden sentences' in pdfs (and it would be ideal if we could also do that for items that only have html) ->

    * exporting them in some format out of Zotero (and doing so with metadata but be nice but not absolutely necessary) ->

    * feeding those sentences as 'seeds' in some NLP-tool and letting NLP do its magic, in close collaboration and iteration with the researchers, in identifying relevant text spans (with a probability score - how likely is this sentence to be relevant?) in the entire corpus

    Hey - you asked for it!!! :)))

  • emilianoeheyns
    If those notes present themselves as objects with itemType "note", the Collected Notes exporter from BBT will export them. If not, open an issue on the BBT github tracker and I'll get you sorted.
  • sdspieg
    Thanks! We'll look into this
  • joycekwc
    Oh wow - thanks @dstillman, the new 'Add note' button in Word is amazing. That is going to revolutionise my workflow!
Sign In or Register to comment.