Got a few spam messages in moderation. But it still puts all the normal messages in the queue as well.
2015.09.16 new accounts still end up in the queue a lot, trying to lower suspicious account to 2 hours
I wonder if posts from āsuspicious accountsā always get moderated - so all new users would have this problem.
It would be great to know which setting (keyword found, URL, new user, abuse report, ā¦) moved content into the moderation queue.
2015.09.16 iāve checked the āWhen the spam prevention service flags content as spam, add a moderator message to the content. This message is only visible from the Moderation page and only moderators can see it.ā setting. Maybe it will tell in the notes why a content is marked as spam. Though i doubt.
Iāve been out all day yesterday (hence the late response), didnāt do any moderation.
We have āSpam Link Preventionā enabled, but there are no domains whitelisted. This will probably result in all posts that contain a link (from users with fewer than 500 points) to be marked for moderation.
I have enabled to option āAutomatically update the domain whitelist when content is approved by a moderatorā
This should slowly start to fill the whitelist, at least with the popular domains. Eventually, that should result in fewer false positives.
Instant update: the whitelist was not empty (but the default page doesnāt show any entries until you hit the āallā button). Regardless, automatically adding āapprovedā domains to the whitelist should help.
Mollom reports that it detected 16 spam messages in the last 2 days. It also claims that statistics are not updated real-time.
Thereās a new message in the moderation queue now. The moderator message simply says āFlagged by spam prevention serviceā
Iāll leave the messages in the queue for now. That will allow one of us (you ) to get a ācompleteā picture of what effects the configuration changes have.
By the way: Mollom claims to be self-learning. Hopefully false positives are marked as such back to Mollom, and detection will improve over time.
Guus der Kinderen:
We have āSpam Link Preventionā enabled, but there are no domains whitelisted. This will probably result in all posts that contain a link (from users with fewer than 500 points) to be marked for moderation.
This was done a while ago when we were getting a lot of spam with links. As normal messages rarely contain links, it was an ok workaround (to get a few normal messages in the queue, but filter off all the spam).
I havenāt read that automatic whiltelist update option carefully enough. I think it is a good option which should help when a user is posting links, but still havenāt reached points threshold. Actually 500 is rather high for many users. It was set that high, so spammers wonāt collect points by updating their status or posting dumb comments. As i said before, i have disabled or lowered many points rewards recently. Maybe we can lower this threshold to say 50. Then spammer account would have to say post 50 new discussions, or comments, or get 10 Correct answer rewards before reaching 50 and getting the ability to post links without moderation. Looks like too much work for them. What do you think?
I have seen that message in the moderation and the system note. As i expected, not much information.
Looks like all the normal messages are hitting moderation because of that āsuspicious accountā setting. There was a message today from an account 2 years old and it wasnāt moderated. Maybe we should try disabling it for now?
Itās great that Mollom has a self-learning feature. Though, i think it probably mostly learns spam patterns. Learning normal patterns seems rather hard in my opinion. Btw, where can you find Mollom reports (couldnāt find anything when logged into www.mollom.com)?
After listening to Jive-x 2015.3 update webinar, there will be some enhancements to the Moderation. Though the sound was not very good, so i couldnāt hear exactly, but as i understood there will be options to moderate content divided by language or set an allowed language for content (maybe will be possible to filter some language? like arabic or korean), moderate content by time (last 24 hours, not sure what it is), moderate repetitive content (important for us, maybe it will be easier to reject lots of similar messages).
Also it looks like Jive has a group in its customers community about Korean spam attacks and they share settings that work against it. Need to look for that.
Hereās the private group about anti-spam ideas Group: Spam Management Ideas | Jive Community
I just joined Group: Mollom beta plugin feedback | Jive Community to hopefully get more insight into the plugin. It canāt be too complicated as the Mollom API is not complex.
Btw, where can you find Mollom reports (couldnāt find anything when logged into www.mollom.com)?
Manage Sites, Site Manager, āEdit siteā, āViewā
We can lower the āSpam Link Preventionā limit from 500 to 50. I simply set it to a more or less random value. Actually users usually do not post links but often the discussions contain logs which the address of their admin URL. As we can not allow http*://*:9090 we get some false positives. I did update the domain list manually in the past as there are only a few URLs which are fine. Adding there 192.168.4.5 or 10.5.33.98 makes little sense.
2015.09.17 lowered spam link prevention threshold to 50
LG, thanks for the Mollom hint. I thought about pressing Edit, but then thought ānah, it couldnāt be thereā The graph shows that it now blocks less and lets through more messages. Not much spam so far. So, maybe it is learning actually. Though, we are using Relaxed mode still.
2015.09.17 disabled suspicious user setting and set the mode back to Normal (based on discussion in the Mollom plugin group at Jive, suspicious user setting will put everything into moderation from the account younger than this setting). As like 80% of messages in the forums are from the new accounts it will put a lot of them into moderation. Will see how it deals with the spam without this setting.
It doesnāt look like link domain whitelist works. Just got a message in the queue with a link to xmpp.org, which is in the whiltelist. Unless Mollom replaces all the other settings with its own flagging.
Perhaps you should poke Jive about this?
Was the content blocked by Mollom?
They also have word and URL blacklists which we may populate (no auto sync by Jive/Keyword interceptor) and maybe whitelists.
The Jive URL whitelist we use has likely nothing to do with Mollom - they check likely other things to detect spam.
I forgot to look at the notes of the moderated postā¦
On Mollom, i see Spam Blacklist, Profanity Blacklist and Unwanted Blacklist. No whitelist, or is it Unwanted? It doesnāt have any description. We can try populating that blacklist if something slips through it. Not much spam recently to have an example.