On Sat, 14 Sep 2019 12:53:27 +0200 Henrik Rosenø wrote: > I don't understand how the word 'som' can be the problem. Almost every > email ever sent in Danish contain that word... There are about 140 of > them in my email... >
Som and Somalia wouldn't have caused the FP alone, the match was specifically on " som ’" The "'" character is a unicode right single quotation mark rather than an ordinary ASCII single quote "'". In UTF-8 this is represented by the byte sequence E2 80 99. When the rule was written ISO 8859-1 was the most common character set, and in that E2 represents an 'a' with a circumflex accent. So the rule was matching something that might have been " som â". I wouldn't worry about this as the sub-rule has been gone for almost a year, and most servers update rules daily. Also mail-tester is using score set 1 (net rules on, Bayes off) which is used on a minority of installations. In the other sets the score is small. score DRUGS_MUSCLE 0.001 2.499 0.392 0.164
