Key decision factors

Performance features

Current numbers

	100.00% clean 0.00% virus 0.00% possible spam 0.00% definite spam
	average email distribution
	24 hours | 30 days | 365 days

	100.00% clean 0.00% virus 0.00% possible spam 0.00% definite spam
	average email distribution
	24 hours | 30 days | 365 days

	33.90% clean 1.81% virus 1.16% possible spam 63.12% definite spam
	average email distribution
	24 hours | 30 days | 365 days

The graph above shows the current distribution of clean email, spam and virus. Reload the page to update the graph.

Filtering

spamchek = filtering + management + information

Filtering is a highly critical part of Spamchek. Based on the email headers and content, the filter section(s) determine if an email is spam or if it contains a virus. The outcome of the filter section determines how the email should be processed, i.e. should it be quarantined or not.

•
•
•
•
•
•
cache



•
•
•
virus




•
•
lists





•
•
rules







•
•
bayes




•
•
•
•
•

Conceptually the filter section can be viewed as a pipeline with several independent stages.

Some stages are conclusive (the virus-section for instance decides on its own if an email contains a virus), whereas others are accumulative, i.e. that is they only work in certain combinations. Also, if the outcome of a definitive stage is negative, it may trigger application of other stages. For instance, even if the virus-stage is negative, a second set of accumulative stages will be applied to determine a probability of the email being a virus.

Filter stages:

Virus	The virus stages utilise the ClamAV virusscanner libraries to access the ClamAV database. See the Virus section for more information, or go to the Virus status page to see the current virus statistics.
List-based	Various elements of the email headers and content (IP-addresses, domain-names, etc.) are checked against several lists, typically in the form of DNSBL, i.e. DNS-based blacklists. We maintain our own lists, but we also subscribe to commercial lists and use freely available lists, such as e.g. the ones published by surbl.org and others.
Rule-based	These are relatively simple stages that apply one or more text or pattern rules and then assign a "score" when a match is found. They are almost always accumulative.
Bayesian	Bayesian filtering is a widely used statistical method that we use to classify email and determine a probability of an email being spam. This style of filter is also often referred to as a "learning" or "adaptive" filter. Bayes-style filtering can be a highly effective measure against spammers' frequently changing habits and methods. Bayes-based stages are always accumumulative.