Hey @bonfire@mayel and crew, if you ever feel like putting a spec down to attach academic citations to #ActivityPub objects for your Open Science Network, ping me.
@bonfire@mayel@jonny This is all spitballing, but I think letting people tag academic publications in posts just like they currently tag accounts could be kinda neat. Shouldn't even need to break compatibility with current software either.
That would be great! We anyway want to federate link preview metadata (I see there's https://helge.codeberg.page/fep/fep/8967/ though nobody seems to have implemented it yet) so it'd be great if it was done as an extension of that, so that non-scientific platforms can still show a subset of the preview information at least? And yeah looking to existing schemas to pull in sounds great, maybe we can collaborate on a FEP draft that just points to those and shows some examples?
Currently the UI preview component checks for many different field names for each info, eg: `e(@metadata, "datePublished", nil) || e(@metadata, "publication-date", nil) || e(@metadata, "date", nil) || e(@metadata, "created", "date-time", nil) || e(@metadata, "DC.date", nil) || e(@metadata, "citation_publication_date", nil)` because the data is ingested from multiple sources in different formats, so would be great to instead transform it into a single schema at ingest.
@bonfire I like FEP-8967, but my intuition is that citation information is kind of orthogonal to link previews. It could well make sense for software to emit/consume both, but if I'm federating an academic article like https://fietkau.science/content_interaction_public_displays, do I want to bundle 25 link previews? It seems incongruent to how those are used. More crucially, not every academic citation will have a URL, it might just be an old book with title + author + publisher + year.
ah based on that example we may be talking about different things, so far we've been thinking more of how you would federate and preview the metadata of https://fietkau.science/content_interaction_public_displays (assuming it is a publication or any kind of research object) rather than how you'd federate all the references if that text was published directly on the fediverse... which could be done different but also it could be app to the UI to decide if/how to use/display the link metadata
@trwnh Thanks, I hadn't come across this. Nice and detailed. I like that this can encode the type of citation (discusses, summarizes, critiques, ...), I don't love that it does it with a long list of optional properties on the citation object rather than one property with a value range. Feels like it'd be a pain to parse for plain JSON consumers this way. Will think on it more.
you can reify the citation as a Citation object and then express the type of citation with the hasCitationCharacterization property? see the explainer's first example on "direct form" vs "reified form"
@bonfire A bit, yeah. I'm sure each one has pros and cons for the purpose, but I'm in favor of picking something that's ready for use with linked data. We might be able to avoid defining a new JSON-LD context that way.
schema.org's CreativeWork is a supertype of ScholarlyArticle. Curious why ORCID doesn't use the subtypes even though they have the type information in their own data. (Their JSON schema is altogether different from their JSON-LD I think.)
<img src="https://cdn.masto.host/indiewebsocial/media_attachments/files/115/384/920/775/155/570/original/3fb467ac51feb6a9.png" alt=" [Caption above the panels:]
How Standards Proliferate
(See: A/C chargers, character encodings, instant messaging, etc.)
<pre><code>[A text-only panel.]
Situation:
There are 14 competing standards.
[Cueball and Ponytail stand facing each other.]
Cueball: 14?! Ridiculous! We need to develop one universal standard that covers everyone's use cases.
Ponytail: Yeah!
[Another text-only panel. The word "Soon:" appears in its own box at the upper left of the panel.]
Soon:
Situation:
There are 15 competing standards."/>
</code></pre>
@bonfire FWIW, Encyclia currently fetches plain JSON from ORCID and re-emits as human-readable ActivityPub Notes. ORCID's own types, which are used in (non-LD) JSON payloads, are listed in the second column here: https://info.orcid.org/ufaqs/what-work-types-does-orcid-support/ Dunno if those map cleanly to schema.org types but the palettes are small enough that we can figure it out.
@julian @bonfire@mayel@smallcircles I have only been loosely following but I am a "spec on the table" kinda gal. Draft spec and implementation first, then bikeshedding later.
@bonfire Right – with OSN you're also thinking about academic creation, not only sharing? So letting authors declaratively cite specific sources in their posts could be good, as would searching/filtering for “what else cites this”.
Although tbh, my motivation for now is interop between Encyclia and OSN. 🙂 I want Bonfire to recognize Encyclia posts as what they are (automated metadata summaries of specific academic works) and separate those from human discussions.
With #OpenScience we talk of an entire field that has a lot to fight for, given all that's wrong around academic publishing processes and the strangehold of commercial parties.
Also a field full of people who might design/deliver rock-solid robust & *interoperable* open standard specifications.
There's ample opportunity to start something like ForgeFed, a dedicated protocol extension spec. Working name: #ScienceFed and it can be homed at https://codeberg.org/fediverse