Design And Implementation Of Short Message Classification System Based On Naive Bayesian

Since 2000, China Mobile and China Unicom launched a short message service and its development speed is surprising, released by the State Ministry of information industry data,.At present Chinese short message quantity reached 1304.6 billion, the mobile phone network that is as a symbol of mobile phone SMS(short message service) is expected to become "the fifth media" which affects our daily life directly. However, while people are enjoying this "thumb culture" which brings the convenience, some of the negative effects are becoming prominent at the same time,for example, Some spam messages which contain pornography, reactionary remarks, fraud, intimidation, harassment and other contents go everywhere amok,these is not only against the vital interests of mobile phone users directly, but also affect the stability of society, so the research that monitoring and intercepting of the rubbish short message become an important topic that the mobile operators face.Currently operators often using SMS monitoring technology is mainly real-time filtering mechanism,call detail records(CDR) analysis mechanism, protocol monitoring mechanism and so on.Real-time filtering mechanism applies white list,black list,cache technology, Keyword matching technology and flow threshold technology, these technologies have passive, system without self learning ability,the high cost of labor, elam error rate and so on.In response to above problems, This paper offer a solution that puting intelligent text classification technology into junk SMS monitoring system and realizing it, it can make up for these deficiencies exist in the traditional SMS monitoring technology,there is a stronger learning ability, Higher confidentiality, You can save costs and reduce the error rate at the same time. In this paper, introduced the Machine Learnnig, Data Mining, Database and the Communication technology, it design a segmentation and text classification algorithm by machine learning and statistical analysis,it taps data from large amounts of data and extract useful and potential information. This article takes a perspective from the message sending and receiving principle, analyses the characteristics of spam messages, combine with the existing spam filtering method,the article introduces method which combin monitoring function module with user settings to filter spam messages adding Improved bayesian to Short Message Service Center Short Message Service Centre Short Message Service Centre Short Message Service Centre Short Message Service Centre...