A multi-objective approach for Vietnamese spam detection

In this paper, we propose a multi-objective approach for generating sets of feasible trade-off solutions for the Vietnamese anti-spam system (using SpamAssassin). The two objectives for considering are the Spam Detection Rate (SDR) and False Alarm Rate (FAR).The experiments were conducted based on Vietnamese spam data set through three scenarioswith different numbers of SpamAssassin rules; and we used the non-dominated sorting genetic algorithm(version 2)–NSGA-II for finding the trade-off solutions. The result of each scenario was recorded to compare with the performance of the traditional approach (single objective optimization).According to the statistical results, the new approach not only achieved more efficient results but also created a set of ready-to-use rule scores which supports different levels of the trade-off between SDR and FAR. © Springer International Publishing Switzerland 2014.