I managed to get spamassassin working with james the other day but it is
scoring too low. I thought of configuring dcc and razor for spamassassin
but after reading the 100 steps to do this, thought a whole java
solution might be better.

Has anyone thought of writing mailets for spamato
http://www.spamato.net/
Dcc - Distributed Checksum Clearinghouse (http://www.rhyolite.com/anti-
spam/dcc/)
and/or Razor http://razor.sourceforge.net/ or pyzor (if they use a
different database)

I noticed from a post at
http://www.mail-archive.com/razor-
[EMAIL PROTECTED]/msg02485.html

that there is a java client to razor contained in the GPL Camels Eye
mail client
http://sourceforge.net/project/showfiles.php?group_id=152968

The source code says its part of a spamato GPL mail filter 
downloadable from
http://sourceforge.net/project/showfiles.php?group_id=136033

Spamato states that:
This includes
Bayesianato: This is an implementation of a naïve bayesian text
classifier, described for instance by Paul Graham. It tokenizes each
message into small text chunks (words) and, depending on their
categorization and significance, decides on spam or ham.
      * Comha: The Collaborative Multi-Hash filter calculates several
        hashes (kind of encrypted messages) for an email and compares
        them to registered hashes on a server. If enough hashes match,
        the message is considered spam.
      * Domainator: This filter is a URL-based filter that queries
        Google for URLs/domains of a message. According to the number of
        Google results, messages are either marked as spam or ham.
      * Earlgrey Filter: This collaborative filter checks if the
        URLs/domains in your messages are listed as spam-related on our
        servers. Our trust system takes care that only creditable
        members are able to affect the results.
      * Razor Filter: This is the first Java implementation of Vipul's
        collaborative Razor filter. We support both public filters: The
        Ephemeral hash-based filter compares (hash values of) text parts
        of a message with previously made user votes in the Razor
        network. The Whiplash URL-based algorithm is somehow similar to
        our Earlgrey filter, but also works with the Razor network. A
        proprietary trust system is employed to ensure maximum
        reliability.
      * Ruleminator: The Ruleminator is a rule based filter like
        SpamAssassin in its early days. You can easily adjust this
        filter's capacity by defining your own rules matching your
        personal liking.


I have already sent a message to the spamato developers asking for
details on how  to do the integration.

I would like to know what is the consensus, would it be a good idea,
does it exist already are there better alternatives.

Andy Bailey

http://www.hazlorealidad.com

PS I know this should be really be for the developers list but Im not a
member,


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to