Thread - Nostr Hypermedia

Gigi dergigi.com 10 months ago

👀

Dawn dawn@spatia-arcana.com 10 months ago

You're getting so close to absolutely blowing the lid off of this thing. Good job. 🫂

EchDel 10 months ago

The mud lay thick upon the stones,

liminal 🦠 liminal@gitcitadel.com 10 months ago

First you think about the regular text search capabilities that already exist on nostr - that's already a massive plus over navigating papers. But then you add in vector embeddings for semantic search and we're in a new world of untapped potential...

liminal 🦠

Still very much a demo, but now here it is. AI + Nostr: 1) Text search on the title + author of a note 2) Semantic search on content

Example: "The cat chased the mouse" is semantically similar to "Felines hunt their prey at night" even though they don't have any words in common. What this means is that you don't need an exact match of the text, you just need to type in words you think are related and it will retrieve the K closest events that are "semantically similar" View quoted note →

View quoted note →

liminal 🦠 liminal@gitcitadel.com 9 months ago

Agents crawling nostr through similarly related content, without relying on a centralized db #Alexandria's gonna have that 😉

liminal 🦠

Still very much a demo, but now here it is. AI + Nostr: 1) Text search on the title + author of a note 2) Semantic search on content

Example: "The cat chased the mouse" is semantically similar to "Felines hunt their prey at night" even though they don't have any words in common. What this means is that you don't need an exact match of the text, you just need to type in words you think are related and it will retrieve the K closest events that are "semantically similar" View quoted note →

View quoted note →

2 replies ↓

Replies (5)