Model for optimising the execution of anti-spam filters

Abstract

During last years, the combination of several filtering techniques for the development of anti-spam systems has gained a enormous popularity. However, although the accuracy achieved by these models has increased considerably, its use has entailed the emergence of new challenges such as the need to reduce the excessive use of computational resources, the increase of filtering speed and the adjustment of the weights used for the combination of several filtering techniques. In order to achieve this goal we have been refined several aspects including: (i) the design and development of small technical improvements to increase the overall performance of the filter, (ii) application of genetic algorithms to increase filtering accuracy and (iii) the use of scheduling algorithms to improve filtering throughput.