Research On The Fraud Sms Distinguish Algorithm Of Comprehensive Supervision Platform Based On Concept Lattice

The study of SMS fraud recognition algorithm is important in the process of constructing SMS integrated monitoring platform.The current algorithm focus on the recognition of keywords and repetition of words.Driven by the huge profit,it is stronger in the anti reconnaissance capability of SMS fraud.Therefore,the FSMSIMP put forward higher requirements for the recognition algorithm of SMS fraud.In the literature of SMS fraud monitoring algorithms,machine learning algorithms are very rare.This paper,we introduce the concept lattice theory,and present a new algorithm LRA,which is a concept lattice fraud text mining based on comprehensive evaluation algorithm.Different from most algorithms which analyze data after data cleaning,the two processes are unified together by the LRA.The LRA algorithm first selects the representative,and more concentrated SMS fraud data on age for analysis.The data which is sampled and tested form the concept lattice in the form of background.Secondly,we establish the function relationship between SMS recognition and SMS cost and the grade of the SMS package.And,in most of the case,two times adjacent detection results on the same parts of the same sub system are generally the same,therefore,the same data is cleaned by LRA to improve the recognition efficiency.Finally,the problem of the algorithm efficiency and the threshold selection of LRA algorithm are prospected.In this paper,the background reduction theorem is proved,and the result is valid form background data.On this basis,the fraud SMS assessment can be achieved by the formal background entropy of the LRA algorithm.The higher the formal background entropy of the same sub system,the higher the degree of the similarity of the sub system is the same as the risk coefficient of the former case.In the contrary,if the entropy of the formal background is lower,the two cases of the suspected coefficient have greater gap.So the similarity of the form background entropy and the content of fraud can effectively reflect the suspected coefficient of fraud SMS,so as to improve the identification efficiency of fraud SMS.