A Junk Mail Filtering Method Based on LSA and FSVM

When we apply SVM (support vector machine) method to filter spam, there will be a problem that data sets are too large to be solved in the algorithm. So we present a novel approach which is a method of combination of LSA (latent semantic analysis) and FSVM (fuzzy support vector machine). In the process of data set building, we adopt LSA method to handle the problem which data sets lies in implicit semantic kindred words and yawp words. Meanwhile, the data sets will be decreased by using this method and it improves the training efficiency largely. Experimental results show that this method outperformed the SVM and Nave Byes filter algorithm in aspect of recall rate. In 3 kinds of methods, the function of combination of LSA and FSVM is better than otherpsilas and its ability of classification identifies is best among them. The experiments show the expected results obtained, and the feasibility and advantage of the new method is validated.