the-algorithm
the-algorithm copied to clipboard
feat(tweepcred): do not penalize until 1.0 friend-to-follower ratio
Resubmission of #305. Closes #421.
There's a reason why that pull request was closed. If you'd like to discuss upping the ratio, I suggest you do it in a GitHub issue.
threshFriendsToFollowersRatioUMass
is used in two places in "the algorithm": the first to calculate reputation, and the second to calculate user mass. This pull request only addresses the first. Barring that, there are several calculations made that depend on this number.
For user mass, if a user follows more than 500 users and the ratio of users they follow to users following them exceeds threshFriendsToFollowersRatioUMass
, their mass is divided by the product of constantDivisionFactorGt_threshFriendsToFollowersRatioUMass
and their ratio (friendsToFollowersRatio
) subtracted by threshFriendsToFollowersRatioUMass
.
There's a reason why that pull request was closed. If you'd like to discuss upping the ratio, I suggest you do it in a GitHub issue.
That PR was automatically closed alongside all other PRs. You don't have maintainer permissions over this repository and have no idea why it was closed, and definitely didn't review it. [REDACTED].
You don't have maintainer permissions over this repository and have no idea why it was closed, and definitely didn't review it.
No, I don't, but I would imagine that Twitter engineers are not particularly eager to merge pull requests from outsiders that would significantly affect how the site's anti-spam function operates.
That's an understandable perspective. Why don't you let those engineers review it, and sit down and behave in the meantime?
I have sent in a PR and now the people with ownership over the code will review it. The PR is here merely as a convenience to attach to the issue, I'm sure you know how this works.
lgtm :+1: