20-04-2011, 12:45 PM
Presented by,
Manjunath N
manju_seminar.pptx (Size: 436.97 KB / Downloads: 53)
Real time monitoring and filtering system for mobile SMS
Introduction:
This method implements the mobile SMS real-time monitoring and filtering by combining the Pinyin Fuzzed Keyword Matching Technology
Statistics of SMS sending load
• During the year 2006, the daily sending road is 1.2 billion and among them more than 30% of the SMS are junk messages.
• According to some incomplete statistics, one mobile phone user receives 8 junk messages a week in average.
• Lucent company has developed a product Mlilife Anti-Spam(ASA).The most famous one for monitoring and filtering SMS
Disadvantages of ASA
• Real-time filtering of the monitoring center impacts the processing speed of the SMSC(Short Message Service Center) and there are no corresponding technical measures to resolve it.
• Comparing the misjudgement of legal SMS and junk SMS, the former brings more loss to the customers and more trouble to the operators.
• The auto-generating technology of the keyword dictionary is absent.
Communication Between SMSC and SMMC
• Block Diagram for SMMC and SMSC
• Block Diagram of SMMC
• Bayesian Learning
• MD5 Hash Algorithm
• MD5 Hash Algorithm
• Pinyin Fuzzed Keyword Matching Algorithm
Overview
• We release the legal SMS, intercept the junk SMS and add operator’s signal language to doubtful SMS.
• Running the real-time monitoring and filtering on the multi-core hardware platform, the strategy above gives consideration to both of the users and the operators, reduces the misjudgment rate of junk SMS and improves the processing speed.
Conclusion
• For the implementation, the system should use a multi-core software platform for it.
• The use of the Pinyin Fuzzed Keyword Matching Technology to get the pinyin keywords and the junk-degree of SMS, can eliminate the interference information of junk SMS effectively.
• That adjusting the users’ credit-grade dynamically according to the behavior character of junk SMS and then combining it with the junk-degree gotten above to give a comprehensive evaluation of SMS-class, enhances the interception rate and classification right rate
• Proposing a new strategy for SMS classification and process which divides SMS into legal SMS, doubtful SMS and junk SMS, gives consideration to both interests of the customers and telecom company by setting legal SMS free, intercepting junk SMS and adding signal language to doubtful SMS.
• Adopting the Bayesian offline learning for SMS sample to create the keyword dictionary for SMS.