Re: [website-dev] Managing Wiki spam - another thing we can do to limit the spam

On 10/25/10 14:33, TJ Frazier wrote:
> On 10/25/2010 06:31, Clayton wrote:
>>
>> The more complete we make this list of blacklisted URLs and keywords,
>> the harder it is for the spammers to pollute the Wiki, and the less work
>> we have to do long term to manage the spam.
>>
>> C.
> 
> If the regex stuff is sufficiently context-aware that it knows when it's 
> checking a URL (or the display-text for a URL), then it probably should 
> be stricter there, and less strict elsewhere.
> 
> It should be very helpful if there is any way that the "save blocked" 
> page could keep an access log, or leave a message on Recent Changes, or 
> some such. A new user who runs into that is probably using H.P.'s 
> Marauder Map: "I solemnly swear that I am up to no good."
> 
> [1] 
> http://wiki.services.openoffice.org/wiki/User_talk:TJFrazier#Just_Testing


Right now the RegEx is just a piped list of words.  It uses the $wgSpamRegex 
variable: http://www.mediawiki.org/wiki/Manual:$wgSpamRegex  It's currently only 
using a custom spammy word list (based on common OOoWiki spam content.  I have 
not added extras like overflow, zero height, blocking href, the case insensitive 
switch etc.  I'm also trying to be "smart" about the words added so we dont' get 
false positives from the list.

I tweaked it a bit to block you from adding the URL you have on your test 
page... I can't block or blacklist the word "essay" alone... but it should be 
blockable in certain specific combinations.

I also tweaked the 
http://wiki.services.openoffice.org/wiki/MediaWiki:Spamprotectionmatch page to 
remove the clue to what word was the problem... no sense giving the answer away 
to the spammers.

Try your tests again... you should be blocked from saving on the URL and the 
sneaky way you entered "essay writers"  :-)

C.
-- 
Clayton Cornell       [email protected]
OpenOffice.org Documentation Project co-lead