Giovanni Andreani (16/12/04 1:53 pm) said:
>>2. SpamSieve has a white list which contains the names and email
>>addresses of messages which have been marked as good. I have had a few
>>problems with this - specifically, where a good email is sent from (say)
>>"Steve <[EMAIL PROTECTED]>". In this case SpamSieve will whitelist
>>"[EMAIL PROTECTED]" (which is good) and will also whitelist "Steve", which
>>is problemmatic. A few days later, I might get an email from "Steve
>><[EMAIL PROTECTED]>", and SpamSieve will pass this through as a non-
>>spam email because it has whitelisted "Steve". I can resolve the problem
>>by deleting or disabling "Steve" from the white list, but I think that
>>there should be a preference option to control the way in which names are
>>automatically whitelisted. It would be better if only names that are
>>reasonably distinct get whitelisted - e.g. first/second name combinations
>>("Steve Smith").
>>
>>Jeremy
>
>But if one trains SS to learn the difference between the two Steve, won't
>that be sufficient?
Yes. When you get the email from spammer Steve, SpamSieve will give you a
false negative result. If you mark the email as spam, it will disable
"Steve" in the whitelist, and future emails from spammer Steve will be
identified correctly as being spam. Then you have to repeat this with
"Robert", "Chris", "Mark" etc. etc.
SpamSieve could avoid these false negatives by having an option for not
whitelisting simple names.
Jeremy
