Research And Implementation Of A Smart Phone Anti-fraud System Based On Spark+

With the rapid development of science and technology and the flourishing development of communication technology,communicating by smartphones has become part of daily life for most people.At the same time,some companies or individuals use the openness of smartphones to harass or even scam in the SMS service and telephone service for the purpose of economic or political benefits,which causes a great harm to the safety of user's personal information and property.Among them,spam message producers send spam messages or fraud messages to mobile phone users for commercial purposes or other fraud purposes,making mobile phone users unbearable.Therefore,it is important to explore and research an efficient and accurate spam SMS authentication scheme to preserve the normal operation of the SMS business and reduce the risk to users of being defrauded.Besides,telecom operators have listed identification monitoring of harassing phones as one of the issues that need to be addressed.The operator usually identifies the harassing phone in a passive manner.That is,when receiving the complaint about the designated harassing phone,the operator confirms it by calling back,and then intercepts the confirmed harassment call.But it is inefficient and not timely and it cannot solve the harassment phone problem effectively.So it is necessary to intercept fraudulent calls effectively and timely and initiatively to ensure the security of user information.Therefore,based on the investigation and analysis of the spam SMS filtering identification schemes and fraudulent telephone interception schemes at the present stage,this article has made the following specific research:(1)In terms of the problem of low discrimination efficiency and single strategy faced by traditional content-based short message malicious identification methods,we propose the naive Bayes machine learning algorithm to identify malicious messages,which can increase the recognition rate by expanding the training set.By using Spark Streaming to build a real-time stream processing program for machine learning,which realizes second-level recognition of fraud information(2)On the problem of the traditional spam SMS filtering algorithm cannot cope with the processing needs of SMS big data,and cannot effectively identify and process the message,we propose Spark parallel computing framework,which combines big data technology with spam filtering technology.And it takes advantage of high fault tolerance and high throughput of distributed file system to improve data processing capabilities.(3)Facing the vulnerability,existing in the passive interception strategy in the interception of harassment calls at the current stage,of reminding users neither timely nor effectively,we propose to use two-tier authentication to identify malicious phones.On the one hand,it eliminates the negative effects that traditional passive intercption calls may have on users by using the pre-establishment of self-owned number library.On the other hand,it promptly gives corresponding prompts by using the method of recording,uploading,and authenticating the unknown phones in an unknown state.(4)This paper designs and implements a smart phone anti-fraud system based on Spark+and conducts functional tests and performance tests.Results shows that the system has perfect authentication capabilities to malicious SMS and telephone data.