E-Mail Filtration and Classification Based on Variable Weights of the Bayesian Algorithm

Abstract:

The co-occurrence word emphasize the word and word internal relations, so its use can improve shortage from the hypothetical of Bayesian algorithm. To build Token Dictionary, Information Gain algorithm is used to choose Tokens, and Synonyms Dictionary is used to acquire more Tokens. By large amounts of training, the matching scores of Token are counted, according to the matching rate the Tokens that is valuable are selected, and the Token Dictionary is established. The proposed method is used to E-mail classification experiment, the results show that the accuracy of spam filter has a well improvement.

• Periodical: Applied Mechanics and Materials (Volumes 513-517)
• Pages: 2111-2114
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.513-517.2111
• Online since: February 2014
• Authors: Zong Jie Wang*, Yi Liu, Zhong Jian Wang
• Keywords: Co-Occurrence Word, Information Gain, Legitimate E-Mail, Naive Bayesian, Spam

References

[1] Qiu Kening, Guo Qingshun, Zhang Xiaobo. The Research of Personalized Classification E-mail System Based on Agent. Computer Engineering and Application. Vol. 30No. 7, July 2005, pp.176-178.

[2] METSIS V, ANDROUTSOPOULOS I, PALIOURAS G. Spam filtering with Naive Bayes-Which Naive Bayes? [ C] / / Proc of the 2nd Conference on E-mail and AntiSpam( CEAS) . California Mountain View, 2006 : 27- 28.

[3] Y. H. Li and A. K. Jain. Classification of Text Documents. The Computer Journal. Vol. 41(8). 1998: 537-546.

[4] Mitchell TM. Machine Learning[M]. McGraw-Hill. (1997).

[5] Kenneth.W. C and Patrick H. Word Association Norms, Mutual Information and Lexicography. In Proceedings of ACL 27, Vancouver, Canada, 1989. PP: 76-83.

[6] CERNET Computer Emergency Response Team. http: /www. ccert. edu. cn/spam/sa/datasets. htm.