Research And Implementation Of Customer Analysis And Mining Of Abnormal Voice Call Behavior

In recent years,with the rapid development of the Internet,various instant messaging methods such as Weibo and WeChat have been rapidly popularized,but traditional voice calls(fixed and mobile phones)have outstanding advantages such as real-time,privacy and authenticity.Still widely used.However,while voice calls bring great convenience to people,they also facilitate the abnormal behaviors such as telecom fraud,malicious arrears and advertising harassment.To this end,there is an urgent need to provide a means to instantly and accurately discover these abnormal behavior customers from a large customer base,and then take appropriate measures to deal with them.Since the above discovery work must process a large amount of customer data,it requires a lot of manpower and material resources.Therefore,how to use effective algorithms to actively discover customers with abnormal behavior characteristics has become a research hotspot in recent years.This paper firstly analyzes the customer data of the telecom operator and selects the customer's voice call record as the source data to realize the mining of abnormal customers.Secondly,the current common data mining algorithms are studied,analyzed and compared.The clustering and classification algorithms with better performance are selected and studied in detail.After that,relevant research results were applied to the actual system,and a customer analysis and mining system for abnormal voice call behavior was designed and implemented.The main research work of this paper is as follows:(1)The current common data clustering algorithms are studied,analyzed and compared.An improved K-means algorithm AD-Kmeans algorithm)is selected and studied and implemented in detail.(2)The current common data classification algorithms are studied,analyzed and compared.The SVM(Support Vector Machine)algorithm is selected and studied and implemented in detail.In addition,in order to improve the efficiency and accuracy of classification,an implementation scheme combining SVM and AD-Kmeans algorithm in the classification process is proposed.(3)Based on the requirements and algorithm requirements of telecom operators,several behavioral feature models of typical abnormal customers are constructed.(4)Applying the above algorithm to the actual system,designing and implementing the customer analysis and mining system for abnormal voice call behavior.The system is based on the J2 EE layered architecture and uses SSM framework to build and develop.(5)The above system was tested based on the real voice call record of the telecom operator.