Análise de técnicas de baixo custo computacional no combate ao spam

Abstract

Unsolicited electronic messages, known as spam, cause several damages to the service providers and to final users. The use of techniques of low computational cost in spam filtering, mainly in servers with many user accounts, is recommended, because they don't compromise the performance of the service. The objective of This paper aims to evaluate the efficiency of three techniques: greylist, body and header message analysis and blacklist. These techniques of low computational cost had been applied in a mail server implemented with free software. Each technique or combination of techniques had been applied by a period of one week. Satisfactory results have been achived. The efficiency reached in unsolicited messages filtering was up to 94,19%. This tax suggests an alternative to the techniques that need high processing to be executed.