(2018-11-12, 01:48 PM)ArchPrime Wrote: [ -> ] (2018-11-07, 10:40 PM)Euan T Wrote: [ -> ]Purge spammer is showing up for most normal posts, but isn't showing up for posts from guests (and by extension, in this case, posts in the moderation queue which are all currently from guests).
I'm not sure if this is intended behaviour off the top of my head, and will have to look into it further tomorrow night.
Did you discover anything?
It looks like purge spammer isn't implemented at all for guests due to it banning users (and guests cannot be banned as such, since they're guests). I'm not sure what we should do about that - whether we should extend it to handle guests by removing their posts (and even that may be difficult in the case of guests where anybody can choose to use a username and Its can be dynamic).
Certainly, in my case, all spam seems to come from guests, and frequently the same quest might spam post hundreds of times - particularly if there has been some inadvertent interruption to other antispam measures for a time.
While guest posts may not be visible to anyone but me (as I need to approve them), it is still excruciating for me as administrator/moderator working through page after page of spam posts and manually deleting them all. Too easy to include genuine posts for deletion amongst the rest if you are on the 12th page of spam, with 100 more to go, and you just want it to end.
To avoid purging a guest who happens to pick a previous spammer's username, is there a way to heuristically detect and purge guest spammers based on other factors than just username? (IP address?, posting frequency?, repetition of message content or key words?)
(2018-11-21, 02:49 AM)ArchPrime Wrote: [ -> ]Certainly, in my case, all spam seems to come from guests, and frequently the same quest might spam post hundreds of times - particularly if there has been some inadvertent interruption to other antispam measures for a time.
While guest posts may not be visible to anyone but me (as I need to approve them), it is still excruciating for me as administrator/moderator working through page after page of spam posts and manually deleting them all. Too easy to include genuine posts for deletion amongst the rest if you are on the 12th page of spam, with 100 more to go, and you just want it to end.
To avoid purging a guest who happens to pick a previous spammer's username, is there a way to heuristically detect and purge guest spammers based on other factors than just username? (IP address?, posting frequency?, repetition of message content or key words?)
There's nothing built in at the moment, but IP address should be getting checked against StopForumSpam already. Posting frequency should be covered by the flood settings I believe, but I've not tested that with a guest.
(2018-12-01, 03:12 PM)Euan T Wrote: [ -> ] (2018-11-21, 02:49 AM)ArchPrime Wrote: [ -> ]Certainly, in my case, all spam seems to come from guests, and frequently the same quest might spam post hundreds of times - particularly if there has been some inadvertent interruption to other antispam measures for a time.
While guest posts may not be visible to anyone but me (as I need to approve them), it is still excruciating for me as administrator/moderator working through page after page of spam posts and manually deleting them all. Too easy to include genuine posts for deletion amongst the rest if you are on the 12th page of spam, with 100 more to go, and you just want it to end.
To avoid purging a guest who happens to pick a previous spammer's username, is there a way to heuristically detect and purge guest spammers based on other factors than just username? (IP address?, posting frequency?, repetition of message content or key words?)
There's nothing built in at the moment, but IP address should be getting checked against StopForumSpam already. Posting frequency should be covered by the flood settings I believe, but I've not tested that with a guest.
I guess in my mind, PurgeSpammer should act as a bulk cleanup tool for those situations where a spammer (whatever their user status) has made it through the defences one way or another - certainly, that would be my request.
Manual cleanup after any breach is just very painful, and it seems to me that in most cases there is enough keyword similarity between spam posts pushing a given product or service (and/or associated metadata) that some sort of heuristic bulk selection might be possible?
Maybe there could be a smart admin review option that, say, allows posts from trusted users to be bulk deselected from any deletion queue, except where posting frequency is beyond a maximum threshold.
Spam management an prevention is definitely something I want to focus on in a future release (like 1.10), but I'd need to take some time to look at what other systems do and what we could do to improve.
Knowing details of the problems people are facing always helps too! For example, the
Dolphin emulator forums seem to be struggling with a human spammer and they think that having an ability to block posts with URLs in for certain groups would be useful.
(2018-12-04, 01:05 PM)Euan T Wrote: [ -> ]Spam management an prevention is definitely something I want to focus on in a future release (like 1.10), but I'd need to take some time to look at what other systems do and what we could do to improve.
Knowing details of the problems people are facing always helps too! For example, the Dolphin emulator forums seem to be struggling with a human spammer and they think that having an ability to block posts with URLs in for certain groups would be useful.
Cheers for that.
I will definitely watch this space!