Information about User Bans and AI Spam Filter

[ShivZen]

Hi Members,

This is to address the false news on community about ‘Admin’ selectively banning users and to explain how AI SPAM filter works.

For a community as big as Samsung Members, there is no 1 person who is assigned the Admin role, in fact, a team of qualified professionals take up this role. The workings of community are transparent and bound by community brand guidelines- especially for Admins and community managers. For us, each one of you is important, and it is our mission to give you a safe platform to discuss everything Samsung.

Please know that user bans are system generated, and following are some of the reasons you might get banned from community:

1. Use of abuse words in your content- be it images or text.
2. Highly negative sentiment in more than 80% of your content.
3. Sharing harmful links on community, which might put other users’ personal information at risk.
4. Reports against your profile beyond a set threshold.

Kindly know that there are many other reasons, and the information is published in community notices and policies.

Further, before any user gets banned, we send them upto 3 warnings. If you think you didn’t get any such warnings, think again! Here are some reasons you didn’t get the warning messages:

1. The warning email was caught in your email’s spam filter.
2. The user registered via a phone number only.
3. DND is activated on phone number which bars system SMSs.

Also, bans have levels. First ban is usually 5 days long, 2^nd ban is 10 days long, 3^rd ban is a month long, and 4^th ban is permanent. Of course, issuing warnings and bans also depend on the degree of violation of community guidelines on the part of the user(s).

In the recent false cases circulated on community, the users were found in violation of either or all points mentioned from 1-4, and before banning the users, system sent 3 warnings to each profile.

Now addressing the issue where some users claim that their content was removed. Allow me to first explain how an AI filter works:

1. In January, we installed an AI filter to check the spam on community to safeguard our community members because as the community grew in popularity and numbers, we were constantly bombarded by spammers who were posting malicious links and advertisements on community.
2. The AI filter is as good as the data set we feed it. In India, we all use a bunch of languages to express ourselves- Hindi, English, regional, profanity, sarcasm, slang! If you have some background on machine learning, you already know that it is an ongoing process to perfect an AI Spam filter. With such a diverse set, it has been a challenge to train the spam filter. But, here’s the good news J
3. The Spam filter has been successfully taking down advertisements and harmful/abusive content. It is even learning expletives and masking them from content. You’ll be glad to know that the Spam filter is performing at an accuracy of ~ 85%.
4. On top of that, we have a team of human moderators who quality check the output of Spam filter daily and restore any wrongly marked posts/comments/images caught  by the spam filter within 24 to 48 hours.

Now, let me throw some light on what triggers the spam filter to mark your post as spam:

1. Abuse words in your content triggers spam filter. If the abuse word is in the image, that is filtered too. Mostly, the spam filter hides the word and lets the post be but if the ratio of words to abuse words is unhealthy, then the post is taken down.
2. Same post/comment posted on the community beyond a daily threshold limit.
3. Phrases and keywords that are identified as spam by the system basis the past learnings. Please note that these phrases and keywords are maintained is a database which is updated regularly. We all know how creative people can get if they WANT to use expletives on the platform.
   
   The users who are complaining that their posts/comment got deleted because they carry feedback and issues. I would like to request them to revisit their profiles- none of their content which carried feedback/issues has been taken down from platform. Only the content that carried either or all of the above cases- were taken down.

Users that have been banned by the system, violated the community guidelines and we cannot manually restore them. The ban will be lifted by the system when it is due.

This community came into being to share queries/concerns and share knowledge among the Samsung users. Please know that we value your feedback- each one of it- be it negative or positive. Negative feedback is what helps us get better and better, and positive feedback keeps us going!

Here are my requests as a community manager:

1. If your content gets taken down, and if it well within community guidelines, please wait for 24 to 48 hours for our human moderators to restore it.
2. If you receive warnings or face a ban, please rethink your interaction on community with peers.
3. Rest assured that none of the negative feedback is removed from the platform.
4. Please keep the community clean and let reason hold the reins.

Cheers!

[Anonymous]

mamm please block him...this type of language not allowed in the community

[Anonymous]

yes  ban this person 

[Anonymous]

mamm see his language..please block him permanently...it's dangerous for community

[Anonymous]

admin, please listen

[Anonymous]

mamm do something..please see language of this user

[Anonymous]

mamm see his language again....he is using different different user name everytime and spamming

[Anonymous]

mamm see ...https://r2.community.samsung.com/t5/Galaxy-S/feed1save1/td-p/4584092

[Anonymous]

I got banned off Samsung Members no reason I didn't violate no privacy terms because I never talked or uploaded content on Samsung members

[Anonymous]

I was banned for Policy Violation when I wasn't even doing anything, the ban wasn't even lifted or I wasn't even warned that I was being banned but I never did anything to break the policy

[Butchieboy2020]

I received a permanent ban from Samsung Members for mentioning firmware. I believe I must have been reported by a user in the community that was considered "the go to guy" and I kinda stepped on his toes with my knowledge and he didn't like it. I believe I shouldn't have received a lifelong ban for something that a serious Android user knows about and uses often. But nevertheless I was banned without waring.