Data Mining Research, Technology And Its Application In Public Security Work

Data Mining has become a hotspot in the fields of artificial intelligence and database technology in recent years. It has been studied in many fields , but blank in the field of public security. The focal point of this study is how to apply data mining technology to the public security system.Currently, with the high-speed development of society, politic, economy and scientific technology, the modern criminal offence demonstrates the characteristics of the speed , intelligence , high technicalization . According to the whole tendency of the crime, there are five crime peaks since the foundation of our country. The quantity of crime rises obviously every time , especially at the fifth peak which took place from the reform and opening-up. Following each crime tide , new kinds of crime appeared. Economic crime , financial crime , and intellectual crime rise rapidly, which exceed the malignant , primitive and instinctive crime much more. At the same time, the proportion of important cases increase too. So , it becomes an emergency to apply data mining technology to find the new rules and apply them to the daily life of policemen , in order to take measures in time, prevent the crimes in advance , improve the efficiency and the ability of response, manage the criminal offence promptly. Many problems are not well solved in the public security according to the traditional Apriori algorithm and Decision tree ID3 algorithm, such as being not able to produce new criminal offence rules , neglecting some rules related to some harmful cases which occupy less proportion , etc.Under such circumstance, the object of our work is to improve the algorithm , and test it through models. Our work is listed as followed:1. To analyze and discuss the problems that were found when data mining was applied to public security system;2. According to the continuous criminal offence, some sensitive problems were put forward to weigh the association rules and data mining algorithm. At the same time, the traditional Apriori algorithm was also improved. The new algorithm improved the efficiency and was proved to be sensitive to the new criminal offence through models;3. According to the traditional Apriori algorithm, some cases that were important but occupied less proportion were easy to be neglected .The new weighed Apriori algorithm worked well to find rules of such important cases;4. To analyze the problems that was found when the traditional ID3 algorithm was applied to criminal analysis, we put forward the optimized ID3 algorithm based on the expert coefficient a . This offered a new method for decision tree to be used in public security system.