Abstract:

The problem of junk message has become more severe. The flood of junk message has not only greatly disturbed people’s life and also endangered public security and social stability. Therefore, the research of accurate and intelligent filter of junk message is of great significance. The research of existing filtration methods indicates that their implement has some shortcoming. The filtration methods based on black and white list are too simple and brutal. Although, the accuracy of content-based filtration has been improved greatly, their complexity of algorithm usually is cause of operator service network jam. The research indicates that the cloud computing technology has a great advantage in scalability, reliability, cost and other aspects. In particular, the scale of computing power can be made of infinite size in low cost relied on its high-expansion of scale. So the cloud computing is a good platform. Based on this foundation, the essay conducted a careful analysis of algorithm principle of content-based filtration and found that almost all the algorithm of content-based filtration currently used is based on Bayes classification theory. After a detailed study and relevant experiment, found that the content-based filter can be implemented by relying on the cloud computing platform and MapReduce programming model.

Key words: cloud computing, junk message filtering, Hadoop, MapReduce