Okay so I have an anti-spam filter on the blog part of dotTech to protect dotTech from spammers. All comments – that are thought to be spam – are sent to the "spam" folder; I check the spam folder on a regular basis to make sure legit comments were not accidently marked as spam and to delete the spam comments. Lately I have noticed spammers are getting more sophisticated.
For example, I was *almost* tricked into allowing a spam comment. What the spammer did was copy a legit comment that was already posted:
So when I was looking over the spam comments I thought to myself "wait a second, I swear I remember reading this comment before – why is it marked as spam?" I was just about to mark the spam comment as not spam and allow it to be posted when I realized that it was a spammer that was duplicating a real comment. Sheesh – tricky, tricky.
Wait, wouldn't that not really do anything? I mean, you have names-linking-to-websites turned off.
That is exactly one of the reasons why I have it turned off – I can't sit here and ensure that every person that makes a comment is linking to a legit website.
However, to answer your question, yes it would do "something" if I approved the comment. You see the spam filters uses bayesian analysis. In other words, it "learns as it goes". Every spam comment that I mark as "not spam" is recorded as a false positive by the spam filter. So, the next time a comment – that may be spam – is made by the same person, that comment is more likely to be approved by the spam filter as opposed to if I didn't approve the first comment.
Most Users Ever Online: 253
Currently Browsing this Page:
Guest Posters: 9
Newest Members: HOGCALLER, raj08871, will56678, xmlld, vishnu, haseen
Administrators: Ashraf (1741), Locutus (1886), amnesia (270)