Patrick Butts wrote: > If you have any suggestions, I'd like to hear them. It's a home mail > server so we don't send many out. I've been saving a lot of spam > because spamwall tended to lean towards everything being not spam > initially, so I showed it what my spam was. I only have about 300 hams > saved when I started with a little over 5000 spams which I put into > their respective directories and rebuilt the database on day one.
You don't *have* to have a normality of one. Its just something to look at if you are having Bayesian issues. As I said in the previous message: > That's the idea, or the assumption. The goal is to have both sides of > the corpus full of messages and thus the database full of words. I > don't know if we have any actual statistics or supporting information if > that means the normaility should actually be 1 - but I have to admit > that (without being a Bayesian mathematician) I assume it should be, or > relatively close. That's a goal of a fully running, fully populated Bayesian database. But of course you cant just pull email out of thin air. You need email going in and out to "build" your Bayesian database with. If you get more spam than ham and it is taking you a long time to fill the ham, then there isn't much you can do about that. Something that I would recommend to help off-set the amount of spam needless collected, is to use various logging options to discard spam instead of collecting it. Options 6 and 7 are listed as recommended choices for certain "Collecting/File Paths" for this reason. If you aren't, you should use these recommendations. But other than that, there really isn't much you can except send more mail to bolster the ham side of the equation. Other than capabilities of various pattern matching against certain parts of a message, the power of ASSP is focused between your whitelist and your corpus. Whitelisted messages will come through and continue to grow the ham corpus, so you should not have to worry about the offset so much for a home server setup. Real people will know when they encountered a Bayesian rejecting prior to being whitelisted. Non-real people senders will need to be white/red/noprocessing-listed ahead of time, or you may need to monitor the [sendAllSpam] mailbox. <./,DanaInfo=parcel.massbar.org,Port=55555+#sendAllSpam> ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Assp-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/assp-user
