Hello Alexander, On Tue, 7 Dec 2004 00:24:43 +0100 (07.12.2004 4:24 my local time), received Tuesday, December 7, 2004 at 12:49:27 +0500, you wrote about "We need your help with BayesIt", at least in part:
ASK> "localization" option where I'm asked to build a translation table in the ASK> style of ä=a (or the other way around?), I've never seen this in any other ASK> Bayes filter (SpamPal's Bayesian, PopFile, K9 - I've used them all), yet ASK> they've all worked quite good. Can anyone explain what this function is ASK> about? It allows detect text-degradation and -classify tokens correctly - store only one variation of token instead all possibilities How it works... As you probably know, there are chars in some alphabets, which seems as symbol from another (or "almost as...") - or some other char from same alphabet Replacement game was very (now less) popular in spam, because it can (can't really long) fool bayes-filter token "Viagra" not identical to "V1ägrä" (for bayes-filters), but it still readable and understandable for human reading. On creating pairs you give BIT possibility test not only found token, but also all possible variation of writing and increase (in ideal) detection quality -- Best regards, Alexander Leschinsky Powered by • The Bat! 3.0.2.10 • POP3 Catcher 2.0.923.1620 • MyMacros 1.11a • AnotherMacros 0.3.21 /24ED1B1E0/ • Universal Macro eXtender 2.1.1074 rc3 Weakened by Windows XP 5.1.2600 ________________________________________________________ Current beta is 3.0.2.10 | 'Using TBBETA' information: http://www.silverstones.com/thebat/TBUDLInfo.html IMPORTANT: To register as a Beta tester, use this link first - http://www.ritlabs.com/en/partners/testers/
