I have a prototype of a project, Brainstorm, which calculates personalized trust metrics and makes them easily accessible by clients throughout nostr. The prototype uses neo4j to keep track of NostrUser nodes connected by follows, mutes and reports, but we’d like to build a general purpose graph database nostr relay capable of managing all event types, with node types that would include User and Event at a minimum; maybe also Relay, Client, etc. We’d like input from all corners.
First question: is neo4j the best choice? Are there other dbs like GunDB we should consider?
Second question: how can we come together as a community and build it? nostr:npub10mtatsat7ph6rsq0w8u8npt8d86x4jfr2nqjnvld2439q6f8ugqq0x27hf is working on one, maybe there are others who can contribute.
Want to set up a call with the nostr:npub1healthsx3swcgtknff7zwpg8aj2q7h49zecul5rz490f6z2zp59qnfvp8p team and tell us your thoughts?
nostr:npub1gcxzte5zlkncx26j68ez60fzkvtkm9e0vrwdcvsjakxf9mu9qewqlfnj5z
nostr:npub1t8pwzkkhhs94e9acgw9jwca9csyl7a4tnpdttu05039um5j7d6xs72gkvf
nostr:npub1hqaz3dlyuhfqhktqchawke39l92jj9nt30dsgh2zvd9z7dv3j3gqpkt56s
Would be great if this would publish Nostr events that clients could query. It could be sharded in multiple replaceable events, e.g. 100 or 1000 "p" tags in each event. So to get the WoT score for a bunch (or one) pubkey, it would require only a quick query and returned a single event with the data.
These events could potentially be broadcasted to wide amount of relays.
Currently, the Brainstorm prototype publishes personalized trust metrics as kind 30282 events as per the Trusted Assertions NIP. Each event publishes one event with multiple trust metrics (personalized GrapeRank, personalized PageRank, hops, verified followers count, and a few more) for one pubkey, as seen from the perspective of another pubkey’s WoT.
Calculating the scores is the hard part. Publishing them is relatively easy. It would not be difficult to publish a modified Trusted Assertion that contains scores for a list of pubkeys, not just one. Maybe a modification of a NIP-51 list.
*30382
That's fantastic! Thank you!
I implemented it in nostr:nprofile1qqsdr0fnxvmn8hxyz8cwazfm8zu9yt7qmc38ll69nkvsgn8dnej4sxckxm0xe and did a screen recording of it:
https://mibo.eu.nostria.app/c99cd0ed6cb8f4c4947c9dcd0a5124a6067b0ad25efaf28bb04a98f9be727d86.mp4
You just implemented personalized trust metrics … in less than an hour??!?!?
Trusted Assertions ftw!!!
🔥 👏🏻 💪🏻 🍻
nostr:npub1gcxzte5zlkncx26j68ez60fzkvtkm9e0vrwdcvsjakxf9mu9qewqlfnj5z
nostr:npub1hqaz3dlyuhfqhktqchawke39l92jj9nt30dsgh2zvd9z7dv3j3gqpkt56s
nostr:npub1t8pwzkkhhs94e9acgw9jwca9csyl7a4tnpdttu05039um5j7d6xs72gkvf
nostr:npub1clk6vc9xhjp8q5cws262wuf2eh4zuvwupft03hy4ttqqnm7e0jrq3upup9
Whose perspective are you using for the scores?
It's NIP-85, it's not personalized by service provider based. Hopefully if there is a provider or specification for getting from user perspective, I'll add that.
NIP-85 allows scores to be personalized. If Alice is on nostr:npub16x7nxvehx0wvgy0sa6ynkw9c2ghuph3z0ll5t8veq3xwm8n9tqds6ka44x, you check to see whether she’s authored a kind 10040 note. If so, it tells you how to find 30382 events with metrics personalized to her perspective (by filtering on the author she trusts to publish her metrics). If she hasn’t authored 10040, you can use a default perspective, which could be your own user account, or could be nostr:npub16x7nxvehx0wvgy0sa6ynkw9c2ghuph3z0ll5t8veq3xwm8n9tqds6ka44x's npub. To set up default Trusted Assertions, go to straycat.brainstorm.social, sign in with nostr:npub16x7nxvehx0wvgy0sa6ynkw9c2ghuph3z0ll5t8veq3xwm8n9tqds6ka44x, sign up, and click yes to activate Trusted Assertions, which broadcasts your 10040 event. I’ll let you know once all the 30382 events are published.
Thanks for explaining. Right now it's just a local setting (the relay), I will write an 10040 later on, but I doubt I'll support full customizability to it, especially the specification allows multiple relays of same rank type. I don't know what users with multiple relays would expect, if all ranks should be shown individually for reach relay, or if the first one should be picked and used.
Waste of time build a self hostable forum that can run locally from an android app.
I still figure it out how a GrapeRank algorithm can be used to calculate a trust score 🤔
