Microbrew MicroSieve is designed to be a high speed spam filter for USENET news. It searches for duplicate and large articles, spambots, and binaries in non-binary groups.
After using some other filter programs written in Perl, it was clear that there needed to be a faster way to filter spam for today's high-bandwidth news systems (greater than 200GB/day), and thus MicroSieve was born. As with all Microbrew products, this software is released under the GNU General Public License.
Given the large amount of spam riding around in major USENET news systems, a spam filter has to be very fast and remain effective. MicroSieve has easily achieved 5000 articles per minute while only imposing a 2% - 10% CPU load on a Sun Ultra Enterprise 2 (one 300MHz CPU) while checking for all of the following:
Large Articles - Rejects articles above a user definable size.
Spambots - Removes any articles sent by popular spambots.
Binaries in Non-binary groups - Only allows binaries to be posted to binary groups, keeping non-binary groups free of clutter.
Path Header Auto-accept - Allows you to set up auto-accept keyed on the path header so that you don't have to keep filtering articles you already filtered on another server.
Path Header Auto-Blackhole - Allows you to set up and auto-reject based on path-header to lock out problem sites.
Duplicate Article History Checking - Allows filtering of duplicate articles based on a custom data-dependent hashing algorithm.
Maximum Cross Post Limit - Sets a limit on the number of newsgroups that an article can be posted to. This is effective since articles that span many newsgroups tend to be junk anyway.
User Supplied Filters - Allows the administrators of news systems to supply their own filters without having to delve into the code of MicroSieve.
What's New in This Release:
· No changes to the 0.3.1 Beta code, just did some extensive testing.
· Updated some old documentation.