Nostr Web Client

1987 is just to have a vector attached to some event, like a label. For a full network, you need to think relationally, a directed graph would need some metadata like "from" and "to". If you know the structure a priori, you can place it as a tag right with the event/node. Otherwise you'll need an index event or something akin to your directory system, where you have specific events dictating the structure.

MichaelJ 5mo ago 💬 1

I think a new event kind that specifies edges between events could be valuable. Different npubs could specify different sets of edges, creating different "overlays" on the underlying event data.

This could be used to build graph databases. One might use embeddings to determine where to draw connections, and how to weight them. This unlocks semantic navigation journeys embedded in an events context. Think how you can jump from one page to the next on Wikipedia almost endlessly, now apply that flow to just about any Nostr event.

Reply to this note

Please Login to reply.

Discussion

jb55 5mo ago 💬 3

I was thinking just a literal and isomorphic translation of the json canvas spec to nostr. Embed any note type in each node

MichaelJ 5mo ago 💬 1

I'll need to read up on JSON Canvas; it's a new concept to me.

I was thinking about graphs and embeddings like what nostr:npub1m3xdppkd0njmrqe2ma8a6ys39zvgp5k8u22mev8xsnqp4nh80srqhqa5sf and nostr:npub1l5sga6xg72phsz5422ykujprejwud075ggrr3z2hwyrfgr7eylqstegx9z mentioned.

liminal 🦠 5mo ago 💬 1

Adding edge events would add a lot, because we're already visualizing 30040/30041 with a star topology. An edge event pointing to a 30040 event means you're connecting to notes that have already been classified together, meaning you can collapse a bunch of nodes into a single one. You can even make the edge events point to tags, or pubkeys which actually would be a fuzzy graph structure, because you don't know how many connections you'll have. 👀

https://easy-peasy.ai/ai-image-generator/images/hand-drawn-node-link-visualization-undirected-graph-10-nodes

Silberengel 5mo ago 💬 1

Well, I guess that would be 1988, nostr:npub149p5act9a5qm9p47elp8w8h3wpwn2d7s2xecw2ygnrxqp4wgsklq9g722q .

Niel Liesmons 5mo ago 💬 1

What's 1988?

Silberengel 5mo ago 💬 1

Your new graph Kind. Gave it the next number, so that it's easy to remember.

Niel Liesmons 5mo ago 💬 1

Aha, so best to use replaceable. Gotcha!

Thanks :prayinghands:

Silberengel 5mo ago 💬 1

Best to use non-replaceable.

Silberengel 5mo ago 💬 1

Replaceable is 20k-30k

Silberengel 5mo ago

30k-40k, sorry.

Silberengel 5mo ago

Yeah, I also don't know canvas, but it looks cool. nostr:npub1xtscya34g58tk0z605fvr788k263gsu6cy9x0mhnm87echrgufzsevkk5s said he'd provide the basis for that. We'll probably understand it better, if we can see a spec or something.

Niel Liesmons 5mo ago

Just checked JSON canvas. It's neat. But the dims and x and y positions are a pain to interop on. (if you cannot specify the formating of the nodes, that is, as with #hypermedia).

But we can totally copy the Edge concept from it.

And then you can still easily import your obsidian boards. While making it an order of magnitude easier to implement this content type.