This may have been asked already.....I've looked but either it's not been asked or I'm not search for the correct keyword combos...so here goes.

In our organization we're using SpamFilter strictly for "tag and deliver" functions, however it seems that many legitimate emails will get tagged as spam if they are not specificly on a white-list filter somewhere.

I'm renaming the corpus folder as we speak, but wanted to find out if I can expect any accuracy at all out of the bayesian filter if we're not using the quarantine fuction of spamfilter....because by not using it I cannot force a delivery for the filter to "learn" from.

I'm hoping from the whitelist/blacklist additions I've done over the last six weeks that the bayesian filter will be more accurate the second time around.....can anyone give any input to this??

Thanks!