Vahric MUHTARYAN wrote: > I think that DCC and RAZOR will improve spam finding , right > ? at this moment I know that querying will take a time but How many > spamcleanhouse we should use ?
That's up to you. I run both DCC and razor with no trouble. > - As for me DCC / RAZOR can be better thing then bayes rules > then why people fighting for setting bayes , instead of this informing > spams to DCC/Pyzor and Razor better I guess ? I don't know about that. I find both tools to be significantly less accurate than a well trained bayes system. Particularly DCC. > - I know that there was a DCC server , any possibility to > install Razor2 server ! No.. Razor2's server side is 100% pure commercial closed-source and only operated by cloudmark. You could possibly license it from cloudmark, but unless you're REALLY huge (ie: AOL sized) and offered a large sum of money (a few million USD) I don't think they'd consider it. They're really not in the business of sharing their server side. Also there are restrictions on running a DCC server. You must participate in the global flood to run your own unless you get a commercial license. See the DCC license for the full details. > - in DCC docs said that DCC mark mails like a SPAM , any > possibility to adding point to spamassassin because if it's sure then > makine spam check with spamassassin will be unnecessary ! I'd not trust DCC's markings any further than I can throw it. DCC marks everything posted to bugtraq and a large number of other technical mailings. Remember.. DCC is NOT a spam detector, it's a "bulk" detector. Any large volume mailing should theoretically be tagged by it regardless of source or spam vs nonspam. You can fix a lot of that with DCC whitelists, but you can't be sure you have them all covered. Let SA use DCC as a scored add-on and be done with it, it's much safer than explicitly trusting DCC.
