Just been able to test it out, interesting to see really! And looks quite accurate as well, well done!
Just curious, what metrics are used to evaluate similarity ?
And is level of connectivity measured by public posts, or does it account for private messages (messenger on Facebook for example) as well?
Looking forward to discover more of your projects @akolonin !
Thank you @lauraorigin! In Facebook and Google+ analysis involves only information accessible via public API-s of those services, which limits only communications (posts, comments and likes/votes) in home pages of the themselves users and does not count groups or home pages of the other users. In Steemit and Golos, almost everything is counted (we don't handle reposts for the moment but we are working on this). Also we are planning to do sentiment extraction everywhere in the close future. Here you can find more detailed information on what we are doing https://aigents.com/papers/2017/Assessment-of-personal-environments-in-social-networks-2017-IEEE.pdf . Shortly, similarity is calculated based on words that people are using but we are going to change it to use tags (if present) too. Most frequent updates are happening on https://www.facebook.com/aigents . Thank you one more time :-)