--enable-signature-headers
This option will cause the DSPAM signature to be written to the message
header instead of body.
IMPORTANT: This option requires that all users either bounce their messages
into DSPAM, forward as an attachment, or implement some macro that will
retain the X-DSPAM-Signature header, which will NORMALLY BE DROPPED by
standard forwarding.
--enable-webmail
The webmail switch is designed for systems where the original message
remains server side and can therefore be presented in pristine format for
retraining. This option will cause DSPAM to cease all writing of
signatures and DSPAM headers to the message, and deliver the message in as
pristine format as possible. This mode REQUIRES that the original message
in its pristine format (as of delivery) be presented for retraining, as in
the case of webmail or other applications where the message is actually
kept server-side during reading, and is preserved. DO NOT use this switch 字串9
unless the original message can be presented for retraining with the
ORIGINAL HEADERS and NO MODIFICATIONS.
--with-signature-life=DAYS
Specifies the default length (in days) a signature should remain stored on
the server. The default is 14 days. This value should accurately represent
the maximum amount of time a user would need to identify and forward
a missed spam, or mark a false positive. Consider vacations. This can
be changed in calls to dspam_clean on the commandline.
（特征激活）FEATURE ACTIVATION
--enable-neural-networking (EXPERIMENTAL可选)
使中心网络支持可用（参见NEURAL NETWORKING部分）。目前只有mysq_drv 和 pgsql_drv
存储驱动支持该特征，而且也还只是试验性的。
--enable-source-address-tracking
通过syslog把垃圾邮件和正常邮件的源地址记入日志。您可以创建一个包含本地MTA IPs 的DSPAM/meta.whichlist文件，这样就让DSPAM跳到下一个“已收”（'Received'）邮件头。每行一个IP。
也可以用改进了的Blackhole Server写入SBL blacklist文件 (http://www.nuclearelephant.com/projects/sbl/)。 字串2
--enable-spam-subject
预先考虑任何疑似垃圾邮件的邮件头主题部分。有些时候这比X-DSPAM-Result域更有用，因为并不是所有的
邮件客户机都支持带自定义邮件头的邮件规则。
--disable-user-logging
禁止每个用户日志文件的写。禁止后用户不能察看图表或历史日志。
--disable-system-logging
禁止系统日志文件的写。禁止后管理员不能察看图表或历史日志。
FEATURE ACTIVATION
--enable-neural-networking (EXPERIMENTAL)
Enables neural networking support (see the section NEURAL NETWORKING). This
feature is only presently supported by the mysq_drv and pgsql_drv
storage drivers, and is still considered experimental.
--enable-source-address-tracking
Logs the source address of spams and innocent messages via syslog.
You can create a file DSPAM_HOME/mta.whitelist which can contain a list of
local MTA IPs, which will cause DSPAM to skip to the next 'Received' header.
Each IP should be on a new line.
Also writes SBL blacklist files for use with the Streamlined Blackhole
字串8
Server (http://www.nuclearelephant.com/projects/sbl/).
--enable-spam-subject
Prepends [SPAM] to the subject header of any messages suspected to be spam.
This is sometimes more useful than the X-DSPAM-Result field, because not
all mail clients support mail rules with custom headers.
--disable-user-logging
Disables the writing of per-user .log files. Users will not be able to
view graphs or history with this feature disabled.
--disable-system-logging
Disables the writing of the system.log. Admins will not be able to view
graphs or other related information with this feature disabled.
算法规则激活（ALGORITHM ACTIVATION）
默认的已激活算法规则已经非常够用，表现了DSPAM中最彻底测试过的算法。没有必要改动任何选项，除非
您对改变DSPAM的默认方式特别感兴趣。
--disable-traditional-bayesian
禁止传统的Bayesian 算法（默认已激活）。
--disable-alternative-bayesian
禁止Brian Burton 的算法，选择Bayesian 算法。不同之处在于：
-用27个例子代替15个例子
字串4
-在计算中出现过一次以上的记号会取走两个扩展槽。当数据很有限时，这一点比较理想
（默认已激活）
--enable-robinson
可用Robinson的几何平均数测试。不同之处在于：
-窗口型号25取代了15
-联合算法也有区别。参见：
http://radio.weblogs.com/0101454/stories/2002/09/16/spamDetection.html
此算法非常陈旧，不推荐使用于产品成果建立。
ALGORITHM ACTIVATION
The default algorithms enabled are e sufficient, and represent the most
well-tested algorithms in DSPAM. It is not necessary to change any of
these options unless you are interested in altering DSPAM's default behavior.
--disable-traditional-bayesian
Disables the traditional Bayesian algorithm (it is enabled by default).
--disable-alternative-bayesian
Disables Brian Burton's alternative Bayesian algorithm. The differences are:
- 27 Samples are used instead of 15
- Tokens appearing more than once may take up to 2 slots in the