Multi-Resolution Concentration-Based Feature Construction Approach
This chapter presents the multi-resolution concentration (MRC)-based feature construction approach for spam filtering, which depicts a dynamic process of gradual refinement in locating the pathogens by calculating concentrations of detectors on local areas and is considered to be able to extract the position-correlated and process-correlated information from e-mails . After a brief introduction of the background, a generic structure of the MRC model, which mainly contains detector sets construction and MRC calculation, is given and the implementation of the MRC approach for spam filtering is described in detail. Then the weighted MRC (WMRC) approach is presented by considering the different activity levels of detectors in calculation of concentrations. Finally, the experimental validation is given.