Analysis And System Implementation Of Criminal Behavior Based On Data Mining

With our country’s modernization and rapid development of the science and technologies, the social, political, economic, science and technology have developped in a fast manner. As we know, data mining integrates some of the latest technologies, including database techniques, artificial intelligence, information retrieval, statistics, data visualization, object-oriented methods, machine learning, high-performance computing, which is an active research branch. At the same time, the illegal criminal activities with the characteristics of high intelligence, high-tech, dynamic, organizational and professional tendency becomes more and more common, rapid changes in criminal behaviors present several challenges for public security police reform, investigation strategies, the efficiency of law enforcement, crime controlling and prevention strategies, etc. There are many countries and cities that pay more attention on the research investment in crime data mining tools and softwares, and pay more emphasis on crime data collection, analysis and mining, aggressive research as well as the development of crime analysis system.In this paper, based on the current state of larceny case accounted for the vast majority of the cases of multiple usurpation case and Burglaries which are likely to cause serious criminal events, The data mining technology including the association rules mining, classification analysis, clustering analysis, are applied to analyze the burglary crime data, and a easy-to-use operation and intelligent Criminal behavior analysis system is introduced, by which helps users predict the burglary crime events. The main contributions of the thesis are given as follows:(1) Burglary crime data association analysis, using FP-Growth algorithm to generate the required frequent patterns, and then generating association rules from the frequent pattern tree, mining the hidden relationships among data in order to identify the relationship between crimes. For example, at the same time the crime occurred, what kind of person is easily become victims. Strong association rules form the knowledge base can help the relevant law enforcement to analyze the criminal events. The law enforcement officer can make deployment decisions, and prevent the crime events.(2) The classification of criminal behaviors, using decision tree algorithm (i.e., C4.5) to analyze the basic characteristics of criminals, such as:offender education, having expertise, with or without regular employment, economic conditions and the degree of the crime. To mine correlation between all the factors causing criminal events to different degrees, discover the factors which frequently work. Finally, it can assist the relevant departments to develop appropriate policies, for example, strengthening legal education, improving skills education, enhancing public protection measures.(3) The clustering of criminal behavior analysis, using the density based clustering algorithm to analyze the information of criminal events, clustering the data have high similarities. Thus, the criminal behaviors of the same category are with more similarities. This can help investigators identify the potential criminal gangs.