Incorrect Spam Flagging for French Community

  • 0
  • 1
  • Question
  • Updated 1 week ago
Hi there,

Noticed that spam flagging for our French community has gotten much more frequent and not as reliable. Before Finding that a lot of posts are being incorrectly tagged as spam, and since our French community is not as active/not monitored as frequently as English, it poses an issue as the post isn't visible until a moderator sees the flag and manually marks it as not spam - meaning other users can't see the post until this happens, meaning an unnecessary delay in response to the OP. 

Although the feature is very useful, wondering if there's a way create smarter filters so that it can better detect actual spam and not unnecessarily mark other posts. Any thoughts suggestions?
Photo of Ranjan

Ranjan

  • 280 Points 250 badge 2x thumb

Posted 2 weeks ago

  • 0
  • 1
Photo of Ranjan

Ranjan

  • 280 Points 250 badge 2x thumb
Hello?....
Photo of Roy

Roy

  • 15,080 Points 10k badge 2x thumb
It can take a year, Ranjan :-)
Photo of Tashina

Tashina, Community & Education Manager

  • 21,116 Points 20k badge 2x thumb
Hi Ranjan,

Sorry for the delayed reply - I completely missed this post!

The best suggestion that I have would be that, as frustrating as spam is, the spam filtering tool needs some "training" to be the most effective. The more that your team can mark topics and users as spam, the better it will be at catching these topics. It sounds like your team is already doing this, but keep it up and it will help.

While it can slow things down when topics are incorrectly flagged as spam, making sure to mark topics and users that are posting spam should help reduce the number of real topics that end of up in the spam filter.

If any other community managers or teams have suggestions on this and how they manage spam in their communities, I would love to hear them.
Photo of Roy

Roy

  • 15,100 Points 10k badge 2x thumb
We don't know who the false positives are catching - Ranjan, can you comment? - but as I think I have suggested before in a slightly different context, some sort of reputation assessment, which could be based on something as simple as Points, could stop established posters being caught by this.

And new posters with few Points could be tar-pitted; this would have a negligible effect on genuine posters, but would stop bots and spam farms in their tracks.

These techniques would work well in any language, I think, and so have no special difficulty with French versus English.

Indeed, perhaps you already do these or similar things, but they are secret :-)
(Edited)