Research on Dynamic Monitoring and Intelligent Early Warning of Community Correctional Recidivism Risk Based on Multidimensional Data Mining

Abstract Reducing the recidivism possibility of correctional personnel has always been one of the social management goals pursued by punishment, and from this level of understanding, the recidivism risk assessment method for community correctional personnel becomes a social management tool. In order to realize the dynamic monitoring and intelligent warning of the recidivism risk of community corrections, this paper proposes the MApriori algorithm based on Mondrian platform, which mines the association rules on the multidimensional data of the community corrections personnel and obtains the basic characteristics of the recidivism of the community corrections personnel. Meanwhile, a model for early warning of community corrections recidivism based on logistic regression is being constructed to monitor the risk of recidivism in community corrections. Finally, the density clustering (DBSCAN) algorithm was utilized to build a model for predicting criminal behavior with the aim of applying it to the field of recidivism research in community corrections. The results of multidimensional association rule mining showed that low literacy, short sentences, young age, and previous burglary were the main characteristics of recidivism. Meanwhile, the main factors affecting the recidivism of drug-related first-time offenders include seven elements, including gender, stable residence or not, drug history experience, occupation type, and cultural level, among which the higher the recidivism possibility of those who are male, 30-39 years old versus 40-49 years old, live in remote rural areas, have no fixed residence, have a low level of cultural level, have unstable occupations, and have a history of drug abuse.