The Research And Application Of Criminal Data Mining System In Public Security

With the rapid development of data mining technology,the PSB industry have more and more criminal data need to be dealed.So far,there is no complete functional PSB criminal data mining system can be used.In this aritcle,we bulid a fully function PSB data processing system with data mining algorithms,by using the association rule extracting algorithm and incremental cluster algorithm to deal with the criminal data.In the data mining algorithms for cirminal data,compared with the traditional Apriori algorithm,we using the proposed FP-Growth algorithm that extract association rule by FP tree,this will save more computing and storing complexity.As for the traditional cluster algorithms,we do a lot of improvement to make the cluster algorithm higher precision and higher robustness:(1)Using FP-Growth algorithm to replace the tradtional Apriori algorithm,and bulid frequent itemsets and FP-trees for extracting association rules.(2)Using recursive function to extract association rules in the tree structure,it can be faster get the associate rules than other structure,besides,the FP tree structure can avoid the traditional Apriori algorithm to get the defect of redundancy frequent items,reduced the computing and storing complexity.(3)In the incremental cluster algorithm,using atrribute reduction feature to be the leaves points to build a core-tree.Using BIRCH creating K cluster centers,and use these cluster centers as leaves to build a core tree,every cluster belongs to a cluster of core tree's branches.(4)Complete the branches and leaves of the core tree by core data.Compare the new object to every cluster center,and compare the threshold,and if blow threshold,then the object belongs to this cluster,besides,get a new cluster center(5)Using K-means algorithm to process the following procedure for leaves points in the core tree,link every leaves to bulid a link list,and recalculate the leaves data,then update the culster center,this procedure will improve the robustness of the core tree.Based on the improved algorithm,we implement a system of Crinimal Data Mining in PSB.After anaylse the experiments reuslts,we found that our proposed system can extract good features and associaiton rules for Criminal data,these useful features and rules not only can help PSB slove cases,but also can give some advice for preventing crimes of ordinary people.