Alxndr

Spam Filter

I just saw your post at Spam Filter. I let this topic slip off my radar screen. If you are interested in an experiment for http://oberwiki.net/ I would suggest the following:

Modify the SpamBlacklist extension so it can use the Regex located here: wgSpamRegexList.txt (formatted for PCRE here wgSpamRegex.txt) to be maintained on a wiki page like Spam blacklist

Caveat: if you view wgSpamRegex.txt in FireFox browser, the long lines may not display.

The spam filter text is divided into three sections:

markup

Example: display\s*:\s*none

URL/domains that cannot be blocked with keywords

Example: \.info\/?\s+[0-9]+\s*\]

keywords

Example: adult.{0,9}(chat|check|live|movie|porn|toy|video)

The SpamBlacklist extension presently dictates the PCRE pattern matches only domain names. This can be changed to block posts with URLs containing banned keywords or URLs followed by banned keywords.

Examples

[http://www.BannedKeyword.org/ Have A Nice Day]

[http://www.HaveANiceDay.org/124/BannedKeyword Have A Nice Day]

[http://www.HaveANiceDay.org/ buy BannedKeyword ]

I haven't had the time (talent?) to make the changes to the SpamBlacklist extension. Instead, a Ruby program loads the Regex (wgSpamRegex.txt) with a cron job at katrinahelp.info. A kludge for sure. You can see the effects here http://katrinahelp.info/wiki/index.php/Spam_Patrol

A typical result is

09-01-2006 -> 34 days bot spam free. 20 days spot spam free. 22 days vandal free.

Further improvements in the SpamBlacklist extension could allow for thresholds, such as, if no banned markup or banned URL is found then at least 3 banned keywords are needed to block a post. The key to such an improvement requires using PCRE features. Another improvement would be to log all blocks.

You can email me here: wallingconsulting-at-gmail-dot-com

--jwalling 19:47, 1 September 2006 (UTC)

RSS extension licensing

Since you are one of the authors of Extension:RSS i would like to hear your comment about User_talk:Mutante#RSS_Extension_Licensing. What would you license it under? Because they want to package it into Fedora. Greets, Mutante 13:29, 12 July 2009 (UTC)