JOURNAL OF COMPUTERS (JCP)
ISSN : 1796-203X
Volume : 4    Issue : 3    Date : March 2009

Application of Refined LSA and MD5 Algorithms in Spam Filtering
Jingtao Sun, Qiuyu Zhang, and Zhanting Yuan
Page(s): 245-250
Full Text:
PDF (123 KB)


Abstract
The paper proposes a spam filtering method that uses integrated and refined Latent Semantic
Analysis (LSA) and Message-Digest Algorithm 5 (MD5) algorithms to address a series of universal
problems in spam filtering, including remarkably lowered filtering precision and notably unbalanced
filtering efficiency as a result of lack of latent semantic analysis of mail contents. In introducing LSA,
its weighting function is improved by integrating fuzzy membership to improve effectiveness of LSA
in processing mail contents. On top of this, MD5 algorithm is used to generate “E-mail fingerprint”,
thus enabling quick matching and realizing highly efficient and accurate processing of mass-
mailing spam. The result of the simulation experiment testifies effectiveness of the method.

Index Terms
Latent Semantic Analysis, Message-Digest Algorithm 5, Fuzzy Membership, E-mail Fingerprint,
Spam Filtering