How does the new JunkMark SPAM filter work?

[Klemen]

Hello,

I received quite a few questions regarding the new JunkMark(tm) filter, so I decided to write it here for everyone.

JunkMark is hidden from the view of visitors (unlike security number check), you don't even notice it from the outside. In the last week I logged and studied over 200 guestbook SPAM messages and the majority of these messages have a few things in common.

What JunkMark does is check every message that comes through the security number check and searches for certain patterns/characteristics that show the message could be SPAM. Then it calculates the probability of a certain message being spam, ranging from 0 (ok) to 100 (spam). The higher the number the more likely the message is SPAM.

Exactly how it works? It's quite simple actually, but I would rather keep this quiet for obvious reasons; the less spammers know about the filter the less chance they have for fooling it.


AFTER YOU TEST THE NEW VERSION FOR A WEEK OR TWO PLEASE LET ME KNOW (REPLY TO THIS POST) IF IT HELPED DECREASE SPAM EVEN FURTHER AND HOW MUCH!

Best regards,

[Gbook Fan]

Hi Klemen and Henrie.
I appreciate the great work you guys are doing with, among others, the Gbook script.
Just one thing, that could save bandwidth for all users, I have added the 'No Spam' banner, but I uploaded it first to: http://www.netmechanic.com/GIFBot/optimize-graphic.htm
and was able to compress the .gif file from 4.20KBs to 1.40KBs without degrading the image.
I hope by advertising this free 'Gif Bot' service I am not in any violation of the terms for using this forum, but thought it could be useful to others.
I will also let all forum users know I get on, in using the new filters provided by our guru, Klemen.

[Klemen]

Hi,

What that tool does is reduce the nuber of colors in the image. I doubt that reducing one 4kb image to less than 2 kb will help much is saving bandwidth , but in case anyone wants the smaller version here it is (right click and select "Save image as" then upload over the original one):
http://www.phpjunkyard.com/extras/nospam.gif

Also feel free to create your own "nospam" images as I am not much of a graphics designer Feel free to post your images here for others.

Regards

[Rizlaw]

I updated to version 1.35 of GuestBook last week and activated the new Junk Filter with the setting of 60%, but I am still getting junk mail entries written to the guestbook - about one or two every week. About the same amount as I received with Guestbook version 1.33.

The last one was from a website called "Nude Celebs". All of the junk entries appear to have a similar and very simple one line entiy along the lines of:

"Very nice web site. Visit xxxxxx website." or "Well done web site. Visit xxxxxxx website."

Seems this type of language is innocent enough to evade the new filter.

BTW, otherwise, this is an excellent Guestbook. Please keep up the stellar work.

[Rizlaw]

Just got another junkmail to my guestbook, here's the info I received in my email notification of the the new junk entry:


"Hello!

Someone has just signed your guestbook!

Name: Nude Celebs
From: US
E-mail: asdfg@hotmail.com
Website: http://www.***************.net/

Message (without smileys):
nice site thanks and good jod"

[Klemen]

Hi,

Well the spam filter isn't perfect and you can't block all the messages. But I don't think having 1-2 junk messages per week is that of a problem, is it? Some people who have contacted me said they received a few dozen spam messages per week before using GBook. Try disabling the JunkMark filter and security number check for two weeks to see just how much is being blocked.

You can't block all the SPAM, but I think GBook does a good job blocking most of it.

Regards

[Rizlaw]

Hi,

Well the spam filter isn't perfect and you can't block all the messages. But I don't think having 1-2 junk messages per week is that of a problem, is it? Some people who have contacted me said they received a few dozen spam messages per week before using GBook. Try disabling the JunkMark filter and security number check for two weeks to see just how much is being blocked.

You can't block all the SPAM, but I think GBook does a good job blocking most of it.

Regards

I agree. However, would it be possible to add, in a future version of GBook, a Blacklist feature where we can add the email addresses of repetitive spammers and the websites they are soliciting people to view, so that GBook can prevent the spammer from adding his worthless and uninvited post?

[Gbook Fan]

Hi Klemen, after having the same problems lately (as has been already detailed on this thread), I have lowered my 'junkmark' setting to 40:

$settings['junkmark_limit']=40;

Do you think this could help in tackling the spam problem, rather than to just turn it off?

Or am I possibly going about things incorrectly?
Sorry for my ignorance in this matter.

Kudos, to all at PHP Junkyard for your great work.

[Klemen]

Hi,

It could help, but remember - the lower the limit, the more possible it is your GBook will block legit entries.

Some times SPAM is very very hard to detect. I will keep developing anti-spam features for GBook, but I don't think we will ever be able to block all the spam.

[fdnyfish]

how about URL filters?

is there anyway to filter URL's with a list of words for the URL fileter.

I have been getting a few messages pass thru...

Hello!

Someone has just signed your guestbook!

Name: SukaSan
From: SukaSan
E-mail: kuzma.kulevskii@mail.ru
Website: http://www.workswashers.info

Message (without smileys):
Good work, webmaster! Nice site!



Hello!

Someone has just signed your guestbook!

Name: Mick Pup
From: NY
E-mail: cvx@yahoo.com
Website: http://tattoo.vpojw.info

Message (without smileys):
Nice site!

[Klemen]

URLs are being filtered, but how would you filter an URL like http://www.workswashers.info ??

[mr tart]

Just out of curiosity, what happens to an entry by a real user but for whatever reason it is regarded as spam?

Is the poster notified or does the post simply not appear anywhere?
I assume the admin is not notified.

Thanks,

Matt

[Klemen]

Hi,

The admin is not notified (you would get too many notifications when you are hit by many spambots), the message is simply blocked from Gbook and doesnt appear.

Regards

[fdnyfish]

URLs are being filtered, but how would you filter an URL like http://www.workswashers.info ??

**workswashers**

i have a problem with tis website appearing in my guestbook:
Website: http://tattoo.vpojw.info

i would like to be able to filter like this:
**tattoo**

[Klemen]

Do you have the latest version of GBook (1.4)?

You would need to write some extra code to filter URLs any further, FruitBeard gave an example here:
viewtopic.php?p=3245#3245