Spamato A Collaborative Spam Filter System

This diploma thesis had the main goal to develop and implement a collaborative spam filter tool. A group of users should be connected to an anti-spam network allowing them to block spam messages cooperatively. The implemented filter uses a server which accepts spam reports of all connected clients to collect information about newly spread spam mails. For every reported message the server generates an identifier (often referred to as “fingerprint”) that is stored in a database for later lookups. To use the filter, a client needs to generate a corresponding identifier of all incoming messages and check if these fingerprints can be found in the database. If so, the mail was reported as as spam before and can be removed from the mail box. Simplified, a spam mail has to be reported once, to remove it from all users’ mailboxes. The main problem of this approach is the fact that spam mails often contain personalized or randomized parts making it difficult to generate a fingerprint suitable to identify copies of the same advertisement, received by different users. Even though the whole text of a spam mail may be randomized it usually tries to make its recipient visit a linked website. Therefore, our filter uses the domains linked within the message to generate a fingerprint of the mail ignoring the remaining text. To allow future extensions and a higher level of portability, a framework with the name Spamato was developed, which serves as a container for spam filters. Since Spamato offers a standardized mail client interface, a mail program does not need to know about the registered filters to check if a message is spam. The client simply sends the message to the Spamato system, which internally uses all registered spam filters in parallel to test the message. Finally, an add-in for Microsoft Outlook was developed which connects to the local Spamato system. The add-in uses Spamato to spam check incoming mails and also offers a graphical user interface to the filter system. Once these three components had reached a certain stage of development we were able to successfully run a beta test with a small group of users utilizing Spamato to check their everyday email traffic.