I do not know what to do Spam

Question

So I recently went live with iask.nz and was bombarded with obvious fake membership sign ups I put the recaptcha on no real differnce I made it so they have to verify with email still they came 569 so far so i made it that I had to verify each membership. BUT i have no idea what is fake and what isnt! I am wondering if I open the floodgates agin but make it that I have to approve the question and answers?? can i do that?

Answer 1

I think it is better to allow free registration but rather to moderate posting for users with less than X points. So, new users will be able to create posts, they will go to the moderation queue, you check the content and then you realize if is spam or not.

For example, in this site, there are users registering and posting about essay writing services or fitness. Clearly, those are spammers. Usually, they have either a link in the post itself or a link in their user profile that points to their sites. Many times they even have quite complete user profiles with the About filled about their good experience they had with the service they sell.

Comments

Yeah this is the best way I’ve found to prevent spam being visible on the site. However it’s still annoying having so much come in and having to reject it all.

Banning user accounts and IPs hasn’t made much difference. I’ve even added custom filters here to block common spam patterns/URLs, but they just changed the format of the spam.

I’ve been doing research into spam filters like Naive Bayes classifier. It seems to work OK in my tests, but you need a lot of “training” data and most sites wouldn’t have that.

---

Why can't we have a common set of training data?

---

Thanks guys I have re installed signup and made everything 150 points see how that goes

---

@arjunsuresh That may be possible but I don't know how reliable it would be. Firstly it wouldn't work for other languages as Q2A is in English. Secondly, you have to train both sides (spam and non-spam). So a different topic site - say, sports - doesn't have the training data to say that a question about tennis is less likely to be spam.

There may be ways around it, like keeping the spam data, then training on the existing content on the site. The other thing is I haven't tested how efficient it is - adding additional checks and lookups for every post might slow it down somewhat.

---

That is true Scott. But the "spam" content should be same for all sites. Also, the spam users also have some pattern - username, email, IP -- these are mostly different from a typical user. Probably it is good if the spam filter can put such users to the to be moderated list.

---

I have read that a lot of spam is done manually rather than an automated bot. So I have put the following text in the 'Custom message on registration form' field under the Users tab in the Admin Panel. "All spam will be deleted immediately along with the spammer's account." I use the Stop Spam plugin and limit registrations to 1 per hour. I tick the box for 'Email this address when a question is posted' under the Emails tab in Admin Panel. Most of the spam posts are obvious and usually include a link to a website. I just keep a track of questions being posted and delete as necessary. Since I have put the message up on the Registration page warning spammers I've gone from a few every other day to zero.

---

Thanks I have effected some of your suggestions.