Suspect Feature Prediction Algorithm Based On Support Vector Machine And Its Distributed Implementation

With the rapid development of social politics,economy and technology,criminal events are growing at a certain rate,crime is more organized,professional and highly intelligent.China's public security information system is not high degree of information,the lack of intelligent analysis and judgments,and scientific decision-making mechanism,the lack of data from the macro to micro problem-finding means,how to use the data mining technology,give full play to the value and role of large police data,and to police work,improve the efficiency of law enforcement and prevention of criminal activities,has become an urgent need to address the construction of public security information problems.In the big data environment,the public security technology is insufficient,many alternative suspects and the forecast method is relatively backward problem,aiming at these problems,proposes the method of using the support vector machine(SVM)to predict the suspect,improve the detection efficiency.Most of the traditional methods of predicting suspects by regression or classification methods,the possibility of the suspect to judge,which may lead to the possibility of wrong judgment.To solve this problem,this paper predicts the characteristics of the suspect,proposes a novel predictive method based on support vector machine(SVM).First of all,this paper introduces the basic principle of support vector machine,on the basis of this model,the prediction model of suspects' characteristics is proposed,and the effectiveness of the model is verified by experiments,proposes a distributed prediction framework for suspects based on Hadoop.The main results of this paper are as follows:(1)According to the characteristics of the problem and the characteristics of SVM,the support vector machine algorithm is applied to the suspect prediction problem.(2)Propose a prediction model of suspect characteristics.Firstly,the data are preprocessed,and feature selection method based on information gain is used to select the feature.Based on the support vector machine,the prediction model of suspect features is constructed,and the model was validated by experiments.(3)The Hadoop-based distributed suspect feature prediction framework is proposed to solve the problem of mass data suspect's feature prediction.The parallelization of feature selection and the operation of distributed SVM are analyzed,and compared with the SVM of single machine,the efficiency of Hadoop is proved.The results of this paper not only solve the problem of predicting the suspect,but also provide a new way of thinking for the suspect to predict,assist in handling the case and improve the efficiency of handling the case,it has a certain practical significance and reference value.