Economic metric to improve spam detectors

Economic lifting has made email spam a scathing threat to the society due to its related exploits. Many spam detection schemes have been proposed employing the tendency of spam to alter the normal statistical behavior of mail traffic. Threshold tuning of these detectors is still a challenging task. Since, shooting down benign emails as spam (false positive), in pursuit of higher detection rates, can be detrimental. In this paper, we introduce a novel economic metric, based on the underlying spam economic system, to assist detectors in keeping their false positives at bay by associating detection accuracy to the spammer's cost. Hence, the sensitivity of a detector does not need to be tuned all the way up to maximize detection, but enough to make spamming cost unbearable to the spammer. Since, spam is all about making money ultimately. We also show that the statistical features used in our spam detectors can easily differentiate spam from benign and we also show that these features are hard to evade by the spammer. Our evaluation shows the effectiveness of this approach in considerably reducing the false positives for these detectors.