Hey everyone, I've been deep in my literature review phase and absolutely loved using SciSpace to organize and annotate papers. The annotation feature itself is fantastic for my workflow.
However, I've hit a major snag that's really slowing me down. When I export my library with annotations to LaTeX/BibTeX, the annotations don't come through correctly at all. They either get jumbled into the wrong field or are missing entirely. It completely breaks my system, where I rely on having my notes attached to the citations for writing.
I'm not alone in this—I've seen a couple of other mentions scattered around. It seems like the export is optimized for their own platform or maybe plain text, but not for LaTeX/BibTeX users who need structured data.
Has anyone found a workaround for this? I'm currently manually copying my notes from SciSpace into my reference manager, which defeats the purpose of having a centralized annotation system. I'm wondering if there's a script, an intermediary step, or even a different export format that preserves the notes in a way that can be parsed into a `.bib` file. Maybe using the RIS export and converting it?
Really hoping we can crowdsource a solution here. This is such a key feature for academic workflow, and it's a shame it's not working smoothly for LaTeX.
Yeah, I've hit that wall too. The RIS export is a bit of a mess for structured notes. What I ended up doing was writing a quick and dirty Python script that scrapes the annotation data from their web interface directly, using the browser's dev tools to find the API endpoint. It's not elegant, but it gets the raw JSON.
I pipe that into a custom parser that maps it to a `note = {}` field in my BibTeX entries. You'll spend less time maintaining that script than you will manually copying notes. If you're not up for scripting, your manual copy-paste is sadly the "stable" workaround until they fix their exporter.
That manual copying step is precisely where I hit the same friction point. Your suspicion about using the RIS export as an intermediary is the right track, but I've found it to be fundamentally flawed for this use case.
The core issue is that standard RIS or BibTeX don't have dedicated, structured fields for rich annotation data. SciSpace seems to dump annotation text into generic fields like `note` or `abstract` without any delineation between different notes or context. When you then convert that RIS to BibTeX, the data is already corrupted. I tried pandoc and BibUtils conversions with equally poor results.
What ultimately worked, albeit with more setup, was to abandon their export function entirely for the annotations. I use their API, as user423 mentioned, to pull the raw JSON data into a separate notes database. My writing system then references both a clean `.bib` file (exported without annotations) and this separate notes file, linking them by DOI. It's a separation of concerns that actually improved my workflow, but it requires building a small integration.