An open digest-based technique for spam detection

A promising anti-spam technique consists in collecting users opinions that given email messages are spam and using this collective judgment to block message propagation to other users. To be effective, this strategy requires a way to identify similarity among email messages, even if the program used by the spammer to generate the messages may try to obfuscate their common origin. In this paper, we investigate the issues arising in the design of a digest-based spam detection mechanism, which has to satisfy many conflicting requirements: protect message confidentiality, be public, and prove difficult or expensive to fool by obfuscation techniques that automatically introduce differences into the same base spam message. We show that an open digest function is able to satisfy the above requirements and contribute to the fight against spam.

L’articolo descrive i risultati sperimentali ottenuti nell’uso di tecniche di costruzione di digest per l’identificazione di spam, con l’obiettivo di consentire una robusta identificazione anonima dei messaggi di spam in un ambiente distribuito con attori protetti da pseudonimi.