https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6154
Summary: Excessive score for FRT_VALIUM
Product: Spamassassin
Version: 3.2.5
Platform: All
OS/Version: All
Status: NEW
Severity: minor
Priority: P5
Component: Rules
AssignedTo: [email protected]
ReportedBy: [email protected]
Hi! I just got a score of 5.0 for a mail based on the single German word
"vollumfänglich". This is a Gentoo 3.2.5 package pretty much with its default
configuration, but without bayes.
X-Spam-Status: No, score=5.0 required=8.0 tests=FRT_VALIUM1,FRT_VALIUM2,
FUZZY_VLIUM autolearn=disabled version=3.2.5-gr0
X-Spam-Report:
* 1.9 FRT_VALIUM2 BODY: ReplaceTags: Valium (2)
* 0.0 FUZZY_VLIUM BODY: Attempt to obfuscate words in spam
* 3.0 FRT_VALIUM1 BODY: ReplaceTags: Valium
As I do three-way filtering using procmail based on these headers, the mail was
considered "probably spam". The cutoff here is at 5.0 on my system. I guess
other people might already consider this spam. In any case, it seems like way
too much score for a single word which in itself is not spam at all.
I believe you should adjust the default scores for these rules.
There might also be room for improvement of this kind of rules. Things that
come to my mind:
1. If two rules apply to the same word, only take the maximum, not the sum.
2. Have those ReplaceTags rules be somewhat case sensitive: while the letter l
does look like a capital I, it is far away from a lower case i, and should
therefore only be considered in mixed-case or upper-case words.
3. Look for a word end, so you don't match these bogus prefixes.
--
Configure bugmail:
https://issues.apache.org/SpamAssassin/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
