Efficient on-line SPAM filtering for encrypted messages

Spamming is the practice of sending bulk messages through an on-line medium to people connected to it. This could be emails, instant messages or social media messages. Social networking sites that help sending and re-sharing messages became common instantly and their user base, inter-user connectivity are increasing at exorbitant rates. Spammers use such sites to instantly spread messages, which often contain even flavored sensitive information, to a wider community of users. Effective filtering of such messages is challenging for two reasons. Firstly, the filter should operate in real time on large volume of data. Secondly, confidentiality of HAM (not SPAM) messages should not be lost. The secure channel existing between the service provider and the user in social networking sites allows filtering the messages only at the service provider end, after decryption. Hence the filter overhead is high at the server side and privacy of legitimate (HAM) messages are not preserved. We propose a novel distributed filtering scheme which can perform SPAM filtering on secure (SSL encrypted) messages without decryption. The power of distributed processing solves the filter overhead at the server side and provides the required real time filter without loss of privacy of the HAM messages.