Stuart McLachlan
stuart at lexacorp.com.pg
Wed Jul 23 18:35:39 CDT 2003
On 23 Jul 2003 at 13:39, Heenan, Lambert wrote: > After my hasty post I did some research on Bayesian filtering and it would > appear to be the case that either the implementation of the filtering is > bad, as you say, and/or the filtering system was not given a large enough > sample of good and bad emails to work with when it was set up in the first > place. > I had my doubts about whether this is a Bayesian filter so did a web search on "Trend SMEX content filter". All I could find at first was thousands of articles/postings similar to this thread saying what a piece of cr*p the filter was :-) Finally found it as Trendmicros eManager (part of Scanmail). Although they don't say how it works explicitly , it appears to use dictionary based rule matching, not Bayesian logic. Their site search engine doesn't recognise the word "bayesian" on their website and searching for it in their knowledgebase timed out every time. -- Support Pegasus Mail and Mercury. Visit http://www.cafeshops.com/pegasusmail Check out my free software at http://www.lexacorp.com.pg