You are viewing a single comment's thread from:
RE: # @steem-ua: Announcements, Partnerships and Exciting News!
Can we do predictive UA via machine learning and identify users who have a much lower UA than expected from their posting quality? It seems like with your data that ought to be possible.
As a test you could go ahead and pull the upvote activity from the @steem-ua activity log then pull the posts and correlate the content vs the vote percentage to see if you can get a predictive ML algorithm working. Then manually rate the posts for subjective quality (w/o looking at the upvote percent) and compare.
Check out these posts from scipio to get a idea of how it works.
https://steemit.com/utopian-io/@scipio/how-to-solve-spam-on-steem-introducing-userauthority
https://steemit.com/utopian-io/@scipio/userauthority-ua-explanations-applications-and-implications