We had the very first research lexicons call today, to discuss interop & standards for research applications! 🎉
(think @standard.site but for science 🧪)
With @mariaa.bsky.social @wesleyfinck.org @aaronstevenwhite.io @byarielm.fyi @ronentk.me
Reach out if you want to be part of these discussions!
Graphs from Pew Research showing that the Iraq war had surprisingly large support even in 2018, especially among Republicans.
This is surprising until you find out that 44% is about the percentage that thought the use of force in Iraq was the right decision *into 2018*. We just have an extremely committed death cult in the US. www.pewresearch.org/politics/202...
Hey #langsky! Any feedback you might have on these lexicons is appreciated! You can start a discussion at github.com/layers-pub/l.... 🐦🐦
@chive.pub does a mix of these two: for eprints, an author can either store the blob + metadata in their own PDS or in a paper-specific PDS (like @brittanyellich.com mentions); for the KG, there is a community PDS that can be modified by trusted editors (somewhat like @blaine.bsky.social mentions).
I will say actually that when I started the project this one derives from a couple years ago, I was using vanilla JSON schemas.
You could certainly define similar schemas for XML or use vanilla JSON schemas or whatever. From my perspective, one point in favor of ATProto is about the infrastructure it provides for discovery + sharing: defining lexicons (rather than another schema) supports hooking into that infrastructure.
What I’m particularly interested in hearing about are kinds of data you work with and whether they can or can’t be represented using these schemas.
Some thoughts on data representation in the current system in the docs:
docs.layers.pub/guides/
docs.layers.pub/integration/
We're ramping up our alpha, and we'd love for you to join at chive.pub!
Since we launched back in January, we've been adding a ton of fun features; and to introduce them, we're going to be doing a series of deep dives over the next couple weeks on chive.leaflet.pub.
Hey #langsky! Any feedback you might have on these lexicons is appreciated! You can start a discussion at github.com/layers-pub/l.... 🐦🐦
i love data, me too meme
Targeted semantic versioning for linguistic data and metadata!
The agenda for #ATScience is up! Check out the speaker list and full schedule below 👇
The range and quality of proposals we received really exceeded our expectations - thanks to everyone who submitted something!
March 27, Vancouver - can't wait to see you there ✨
Sessions include >
I don’t collect eye movement data anymore (and when I did, it was infant preferential looking data, not reading), so I’m unfortunately not super up on current tech, especially when fine-grained time course and localization of the gaze target is required.
The Grok prompt that was used to design this definitely contained “inspired by the set design of The Man in the High Castle.”
Layers (v0.4.0) now has a full set of XRPC queries with domain-specific filter parameters for every record type, and a major simplification to the way segmentation works.
Github: github.com/layers-pub/l...
Docs: docs.layers.pub
Vibe vs buy: you've never been paying for the code; you're paying for "someone else has thought this problem through in much more depth that I'm willing or able to"
Me once I've exhausted my social battery 10 minutes into every party I've ever been to.
For someone who thinks this, what is lying by omission?
Pan Sonic is on everyone else’s gym rotation too, right? bleep.com/release/2656...
Been tinkering with schemas for sharing language data over ATProto (the technology that underlies bluesky). Trying to support everything from acceptability and inference judgments to interlinear glossing to syntactic parses to eye and neural recordings. Discussion and comments appreciated. 🐦️🐦️
The citations tab for an eprint on Chive showing automatically extracted citations.
The related tab for an eprint on Chive showing automated similarity scoring.
@chive.pub now has automated citation extraction and similarity scoring!
📹 + 🧠 + 📝 = 🔥 First call for MAGMaR 2026, the 2nd workshop on multimodal augmented generation via multimodal retrieval! If #RAG isn't hard enough for you, try multilingually and multimodally. Collocated with
@aclmeeting
in San Diego in July.
nlp.jhu.edu/magmar/
And actually, come to think of it, it is possible for users to have pub.chive.graph.* records. There just aren't currently frontend components that support creating those records in their own PDS. But pub.chive.graph.* is probably how Chive would do user-scoped collections.
An alternative to lists might be typed multidigraphs, which would support richer annotations on the collection. So for instance, while Chive doesn't have user-scoped collections (beyond the user's own eprints), eprints are one node type within its KG, which is such a typed multidigraph.
Re: (a), it is possible to upload an eprint but not be a listed author.
Under (a), the materials would show up on @chive.pub under the @atproto.science profile. Under (b), they would show up on each author's profile on @chive.pub with a link to the @semble.so collection.
Chive doesn't really have a notion of collection. Two possibilities: (a) materials could be uploaded to @atproto.science (but authors wouldn't have edit permissions); or (b) authors could upload their own materials and they could be collected on @semble.so with backlinks on @chive.pub.
Sounds sick.
We also have Jane Chandlee coming right after the class reads her 2017 Morphology paper. link.springer.com/epdf/10.1007...
