I see some similarities with work we did using Neo4j to track follows/unfollows for push notifications. Since the graph was already there, we added a PageRank API too:

🔗 https://followers.nos.social/api/v1/trusted/89ef92b9ebe6dc1e4ea398f6477f227e95429627b0a33dc89b640e137b256be5

It’s only HTTP for now, but I’m considering exposing it on Nostr with the same custom kinds you use for PageRank data 🤔

Code: https://github.com/planetary-social/nos-followers

cc nostr:nprofile1qywhwumn8ghj7mn0wd68yttsw43zuam9d3kx7unyv4ezumn9wsq3yamnwvaz7tmsw4e8qmr9wpskwtn9wvqzqak8r2hr5jglrk0wc37t59lz98x6gyf6pwaku6hpwakhvslznjh6z5205k

Reply to this note

Please Login to reply.

Discussion

yes, it's similar. I've looked into neo4j and memgraph and I wasn't happy with the performance, so I've built my own graph database on top of Redis.

The result is that, instead of computing Pagerank every hour, storing the values and reading them, I can compute Pagerank every time a new follow list has been published.

So it's a dynamic Pagerank based on monte Carlo.

Also, it allows for personalized pagerank.