mplorentz on Nostr: Finally got around to reading this “Navigating the social graph” paper from ...
Finally got around to reading this “Navigating the social graph” paper from
pippellia (npub176p…vgup). I did not know about adjacency matrices. Is anyone representing the follow graph as a matrix in Nostr (or elsewhere)? I would love to learn more about how this works. Right now Nos builds up DAG of the two hops follow graph in memory as-needed. It’s quite expensive in terms of performance and I wonder how we might take advantage of adjacency matrix operations to make it cheaper.
Published at
2024-07-01 22:09:25Event JSON
{
"id": "8fdfe41fa2163ebe3dd5ae3086dc8be07402fb8d124519c195bb5f3290403186",
"pubkey": "d0a1ffb8761b974cec4a3be8cbcb2e96a7090dcf465ffeac839aa4ca20c9a59e",
"created_at": 1719871765,
"kind": 1,
"tags": [
[
"p",
"f683e87035f7ad4f44e0b98cfbd9537e16455a92cd38cefc4cb31db7557f5ef2"
]
],
"content": "Finally got around to reading this “Navigating the social graph” paper from nostr:npub176p7sup477k5738qhxx0hk2n0cty2k5je5uvalzvkvwmw4tltmeqw7vgup. I did not know about adjacency matrices. Is anyone representing the follow graph as a matrix in Nostr (or elsewhere)? I would love to learn more about how this works. Right now Nos builds up DAG of the two hops follow graph in memory as-needed. It’s quite expensive in terms of performance and I wonder how we might take advantage of adjacency matrix operations to make it cheaper.",
"sig": "f5a90c9c3ee361cceca48c8fac5006922e013e4941462209a0854ebd884c6ef632fdae352fe60bb7866255c368d62295efb63d280190733920d47e2c22b34484"
}