Hi all, i wrote some patches for Dspam 3.8.0 that allow user to specify own whitelisted domains via Dspam Web Interface. I think it can be modified a little to whitelist specific u...@domain adresses, but domain based settings works for me fine too, so maybe someone wants to use it. I'll try to add user-based whitelist when i find some time.
Next case is using Dspam with global user classification. I've set
classification group:
group:classification:*globaluser
and trained globaluser with ca. 50K messages. There are some problems
when user's result confidence is small. Globaluser results have
precedence over user result and it leads to discrepancy, for example
message is marked as innocent, dspam writes innocent to logs, but
message header is swapped and shows X-DSPAM-Result: Spam.
Following patch seems to correct this issue (log lines can be dropped of
course) so it would be good, to add it in next Dspam release.
--- src/dspam.c-o 2009-03-30 11:39:22.000000000 +0200
+++ src/dspam.c 2009-03-30 11:41:14.000000000 +0200
@@ -2711,6 +2711,8 @@
memcpy(&CTX->totals, &CTC->totals, sizeof(struct_ds_spam_totals));
free(CTC);
CTX->totals.spam_misclassified--;
+ strncpy(CTX->class, LANG_CLASS_SPAM, sizeof(CTX->class));
+ LOG (LOG_INFO, "Unsure, global user says: spam");
CTX->result = result;
}
@@ -2735,6 +2737,8 @@
memcpy(&CTX->totals, &CTC->totals, sizeof(struct_ds_spam_totals));
free(CTC);
CTX->totals.innocent_misclassified--;
+ strncpy(CTX->class, LANG_CLASS_INNOCENT, sizeof(CTX->class));
+ LOG (LOG_INFO, "Unsure, global user says: innocent");
CTX->result = result;
}
}
Thanks for any suggestions, let's make dspam better ;)
Cheers,
--
Marcin Rzepecki
m.rzepecki(at)iem.pw.edu.pl
dspam-whitelist-patch.tar.gz
Description: Binary data
------------------------------------------------------------------------------
_______________________________________________ Dspam-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dspam-user
