Design And Implementation Of Support Vector Machine Monitoring System Based On Spam SMS

Spam short message is a serious social problem, and it has caused great concern all sectors of society. Spam messages not only interfere the user’s normal work and life, make the user dissatisfaction, but also the spam messages flooding affected the company’s social image.Many users mistakenly believe that sending spam messages with the carrier-related, but in fact spam SMS does not generate revenue for the operators.Study found that the existing monitoring system to intercept the flow of spam messages based on keywords and content strategy to intercept the information security departments to develop automatic imputation spam messages and the formation of "spam messages suspicious library." It manned process in 7 * 24 hour by the customer service center,implement the closure mechanism of "quasi-real-time manual review", manual review "spam suspicious library", restore and release the spam messages that don’t meet the standards,spam messages for compliance with standards with black handle, while feedback keyword library updates. Followed by the keyword update the information security department to monitor feedback based short message gateway exchange ratio value and customer service centers, and analyze the behavior of spammers law, submit maintenance department to dynamically adjust tactics to intercept spam messages and keywords updates. Performance is there, but the system in terms of intercepting spam messages, there are some drawbacks, main containing:1. The filtering process is slow, and can’t guarantee timeliness.2. Manual adjustment of dynamic keyword, time-consuming and error rate is too high.3. It belongs to remedy the situation, "quasi-real-time" instead of "real-time".In this paper, junk SMS monitoring system timeliness lower error bar high,non-real-time defect, designed and implemented a spam messages based on support vector machine, to facilitate real-time statistical analysis of the recombinant intermediate results can reconstruct the originator interception system, and principles for monitoring and interception NE treatment technology to compare the advantages and disadvantages of the system designand architecture, and system analysis of key modules, system design focuses conducted,thereby forming a closed-loop control system integrated spam messages.Through a large number of experiments tested using support vector machine classification algorithm to complete the text short message filtering job classification accuracy improved significantly, ensure accuracy, while greatly improving the social perception of users to solve the existing monitoring system the defects interception performance has been significantly improved.