Installation manual

40
Number of scanned messages before writing them to disk: - While training, the antispam engine will process a
configurable amount of messages before writing the training database to disk. This option determines how many
messages to process before writing to disk. For maximum performance, this number should be as large as possible.
In an unusual case when a program is unexpectedly terminated before buffer has been written to disk, the training
performed since the last disk write will be lost. The buffer is written to disk on normal termination.
Use country data for training - Controls whether country routing information should be considered when training
and scoring messages.
3.3.2.1.3 Rules
Use rules - This option controls whether slower heuristic rules are used. Accuracy can be greatly improved but
more memory is used and it is much slower.
Use rule set extension - Enables the extended rule set.
Use second rule set extension - Enables the second extension to rule set.
Custom Rule weight: - This option allows overriding weights associated with individual rules.
List of downloaded rule files: - This option specifies which rule files are downloaded.
Category weight: - Allows the end user to adjust the weights of categories used in sc18 and in files used in custom
rules list. Category: Name of category, currently limited to SPAM, PHISH, BOUNCE, ADULT, FRAUD, BLANK,
FORWARD and REPLY. This field is case insensitive. Score: Any integer or BLOCK or APPROVE. The weight of rules
matching the corresponding category will be multiplied by the scaling factor to produce a new effective weight.
Custom rules list: - Allows user to specify a custom list of rules (i.e. spam, ham, or phishing words/phrases).
Custom rules files contain phrases in the following format on separate lines: phrase, type, confidence,
caseSensitivity phrase can be any text except commas. Any commas in the phrase should be deleted. type can be
either SPAM, PHISH, BOUNCE, ADULT, or FRAUD. If anything other than those are specified, the TYPE is
automatically assumed to be SPAM. Confidence can be from 1 to 100. If type is SPAM, then 100 indicates a higher
confidence of spamminess. If type is PHISH, then 100 indicates a higher confidence of phishiness. If type is
BOUNCE, then 100 indicates a higher confidence that phrase is related to bounces. A higher confidence is more
likely to impact the final score. A value of 100 is a special case. If type is SPAM, then 100 will score the message as
100. If type is PHISH, then 100 will score the message as 100. If type is BOUNCE, then 100 will score the message
as 100. As always, any whitelist overrides any blacklist. caseSensitivity value of 1 means that the phrase will be case
sensitive; 0 means that the phrase will be case insensitive. Examples:
spamming is fun, SPAM, 100,0
phishing is Phun, PHISH, 90,1
return to sender, BOUNCE, 80,0
The first line means that all variations of "spamming is fun" are considered as SPAM with a confidence of 100. The
phrase is case insensitive. The second line means that all variations of "phishing is phun" are considered as PHISH
with a confidence 90. The phrase is case sensitive. The third line means that all variations of " return to sender " are
considered as BOUNCE with a confidence 80. The phrase is case insensitive.
Clear older rules after their update - The antispam engine, by default, will clean up older rule files from the
configuration directory when a new file is retrieved from the SpamCatcher network. However, some users of the
antispam engine will want to archive older rule files. This can be done by disabling this cleanup feature.
Show notification after successful update of rules
3.3.2.1.3.1 Rule weight
Set rule file indexes and their weight. To add a rule weight, press Add... button. To modify existing, press Edit...
button. To delete, press Remove button.