it gets your last 500 events, and your follows' kind:6s, in chunks of 50 follows and limit:100

it then puts your last 500 events in a bitmap with a window size (30 seconds, which is too short); it's not clustering your activity and it should because I didn't want to take the time to implement it this morning

it goes through all the events that were reposted by your follows and, if takes out any event id that happened during a time you were active

it sorts the remaining event ids by number of times they were reposted by your follows

then fetches them all and removes anything you reposted and anything that was posted during an active time

then it returns 25 random items from the top 100 events by reposted-count score

Reply to this note

Please Login to reply.

Discussion

Interesting, it could be possible to also consider reactions, zaps and the time they are made by follows.

Perhaps a DVM would be the way to implement it, example code:

https://gist.github.com/braydonf/7a86c9b316e6939ae7d8dcb8b1dd601c

More details:

https://github.com/vitorpamplona/amethyst/issues/337