Research On Cloud Computing Technologies For Mass Spam Mail Filtering

Nowadays the problem of spam emails is becoming more and more serious. It not only consumesnetwork resources being a serious threat to network security, but also causes serious distress topeople’s daily lives and poses a great challenge to traditional anti-spam mail filtering technology. Theemergence and development of cloud computing breaks inherent mode and proposes a new type ofdistributed parallel programming model and service application model which provides a newapproach to the topic of anti-spam filtering.The article selects Bayesian mail filtering algorithm as an object of study.After in-depth study ofthe core technology of cloud computing in terms of massive data processing, we improve Bayesianmail filtering algorithm against drawbacks of low efficiency and early training resource-consumingfor traditional distributed Bayesian algorithm. It is the Bayesian mail filter MapReduce model basedHadoop open source cloud architecture.Also it introduces feedback learning mechanisms to adapt tothe constantly updates and changes in the spam, improving efficiency of spam filtration. Experimentalresults show that Bayesian mail filtering MapReduce model can keep good performance of recall rate,checking precision rate and true positive rate. At the same time, it improves efficiency of the filter.After contrasting different roles of email filtering type, we design a SaaS cloud filtering servicemode for mail service provider. It uses mail filtering mode in MDA-side combining with Bayesianspam security filtering MapReduce model which implements in Hadoop platform. The mode containsapplication service layer, cloud filter layer and hardware resource layer. It can provide convenient,customizable and bookable, low-cost, safe, reliable mail filtering capabilities for service users.