Using Data Mining Techniques to Discover the Potential Suspect on alled Detail Record

• Main Authors: Lan cheng syuan, 藍承炫
• Other Authors: 陳垂呈
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/09103995553183379178

碩士 === 南台科技大學 === 資訊管理系 === 93 === Communication technology brings the convenience not only in daily life but also in the work. But the situation that it is abused is worse and worse. The most influence is that gangsters use it as communication tool to do something criminal. In the meanwhile, it also becomes one of the targets that the law officers detect. The telecommunication company records users’ call detail record to count every user’s bill. Through analyzing and studying the users’ call detail record, we can learn the suspects’ relation to friends, living habits, the area of the movement and the probability of crime, even the potential criminals. In the study, we take the information about users call detail record to study the suspects who trend to commit a crime. We provide two methods to find out the potential suspects. One is to look for the suspects. In every user’s call detail record, we count the proportion of frequency that suspects call the suspects who have committed a crime before. When the number adapts the “minimum support of suspects” ,we call them potential suspects who trend to commit a crime. Otherwise, we adopt Association Rules Y→Z. Y is the number of name. Z is the frequency that suspects call the suspects who have committed a crime before. It can show the feature of the trend through Association Rules. We can find out the potential suspects. Tow is to lock certain user. We postulate that the call detail record is D. First, we count the proportion of frequency that the user calls the suspects who trend to commit a crime. When the number adapts the “minimum support of suspects” ,we call him potential suspects who trend to commit a crime. Otherwise, we adopt Association Rules the user→ Z. We can find whether the user is the potential suspects who trend to commit a crime or not through Association Rules. Otherwise, we take the same method to find out the Association Rules Bi→ Z. B is the number of single name. BiD-the suspects who trend to crime. We can find whether the user is the potential suspects who trend to commit a crime or not through Association Rules. The result will provide very useful information so that the law officers can research the suspects who trend to commit a crime and prevent the crime.