The application of LSI truncation in spam filtering dimension reduction

In this paper, the issue of dimension reduction in order to reduce the computational effort for spam filtering is studied, which is particularly important for time-sensitive online spam problem. In particular, to solve this problem, the effects of truncation of the singular value decomposition in latent semantic index are investigated and compared. It is shown that, in the context of spam filtering, a surprisingly large amount of problem reduction is often possible under the proposed method without heavy loss in filter performance.