Co-offending Network And Its Prediction Based On Offender’s Geographic Characters

The rapid development of cities has attracted a large number of migrants,and the reconstruction of social relations and the emergence of subcultural phenomena have led to the co-offending relationship between criminals with different regional characteristics.At present,there is a lack of research on the composition and migration factors of the migrant population about mi-grant crime in specific cities.Moreover,in terms of the influence of regional characteristics of criminals on co-offending relations,the analysis is focused on statistical methods,and there is a lack of research on the selection of co-offending partners and corresponding reasons on the level of regional characteristics of criminals.In view of this,this paper takes Beijing burglary cases data of 2005-2018,for example,to carry out empirical research,and analyzes the regional characteristics and driving factors for non-local criminals.Then,we will research criminal cross-hometown relationship network and its causes in the complex of environment,and construct a prediction model of criminal co-offending relationship network based on regional characteristics.First of all,the paper analyzes the spatial distribution characteristics,influencing factors and trend of migrants’ emigration places by using spatial statistical method,and determines the relationship between the total number of migrants and the number of migrants’ criminals by using correlation analysis,which further explains the impact of large-scale migration and aggregation of migrants on urban crime risk.Secondly,based on the regional characteristics,the author constructs the co-offending relationship network,analyzes the composition and the choice preference of criminals’ regional characteristics,and studies the influencing factors.The results show that the migration areas of migrants in Beijing show obvious imbalance effect and spatial aggregation effect,and the overall spatial distribution aggregation is increasing year by year.Population size,per capita income,education level,population density and transportation time have significant effects on the spatial migration of migrants,and the spatial spillover effects of the geographical culture of neighboring cities on the migration of migrants to Beijing.Large-scale population flow and aggregation to some extent improve the level of urban crime risk.Different areas of co-offending relationship between criminal groups showed a trend of weakening,and with the regional criminal origin number,the more the formation of cross-hometown co-offending rate is lower,in contrast with the regional source criminal quantity is less,is the formation of cross-hometown co-offending rate is higher,reflecting the criminal in cooperative partner selection preferences.The behaviors of criminals participating in joint crimes in different regions are easily affected by the regional cultural background and the spatial distance between the criminals’ belonging places.If the criminals belong to the same geographical cultural area or the spatial distance between the criminals’ belonging places is smaller,the more likely it is to form a joint crime relationship,and the spatial distance has a greater influence.Finally,based on the recommendation algorithm of deep learning,constructing a data set,a prediction model of criminal co-offending relationship based on regional characteristics is established by taking population,education,income,interprovincial distance,cultural area and cooperation situation of different regions as characteristic labels,so as to realize the prediction of criminal co-offending relationship of different regions.Compared with the empirical network,the model is proved to be reliable.The work of this paper can effectively study and judge the joint criminal relationship between the alien criminal groups in the city,which has certain practical application value.