Research On Internal Risk Assessment Of Prison Based On Static Data Of Prisoners

The people in the prison are special.The security hole of the prison will cause social security problems,increase social instability,and lack of efficient management will not be conducive to the transformation of criminals.It is difficult to carry out prison work in an orderly manner.Prison safety management controls the occurrence of prison risk events.Prison risk events can be caused by external factors,such as situational policies,but also from internal factors.Most of the current influential prison hazards come from internal prison risks.Prison internal risks are the focus of prison risk and the most common risk faced by prisons.However,at present,China’s risk assessment is still in its infancy,and the assessment relies too much on the subjective judgment of supervisors and lacks scientific and effective assessment methods.At the same time,China has established a grassroots informatization system and generated a large number of prisoner data.However,at present,China’s analysis of prison data is rarely used.Mining grassroots data and using it for prison risk assessment is an effective qualitative and quantitative method.A combination of assessment methods.This paper is mainly based on the static data of prisoners in a prison,such as archival data,to study the internal risk assessment of prisons.Firstly,this study proposes the concept of prisoner characteristic risk,and divides the risk into four types of risk: violent,psychological,instigating and lacking identification.The four types of characteristic risk of the prisoner are used as the evaluation target.This study aims to address the lack of identifiability of risk assessment results.The prediction of a single risk in the traditional method is converted into a prediction of four specific risk types.The classified features are labeled separately,and then predictive modeling is carried out.The evaluation results can reflect the degree of different types of risks of a prisoner,and it is convenient for managers to carry out targeted management.This paper introduces an improved interval fuzzy VIKOR algorithm to optimize the accuracy of the annotation set.The four types of risk characteristics cannot be quantified by historical data,and comprehensive expert judgment is needed to improve accuracy and authenticity.MADM(multiattribute decision making method)can combine insufficient historical data and expert judgment to calculate risk value by sorting the label set.However,there are two loopholes in the traditional VIKOR.On the one hand,in the prison context,the current indicator weight is subjective,and the weight of subjective judgment indicators will cause uncertainty in the results;on the other hand,the "individual regret value"-The numerical calculation of R in the algorithm is related to the correlation between the indicators.The original algorithm does not consider the correlation between the indicators,so that the calculated value of R is too large,and the disguised form expands the weight of “individual regret”,which leads to the increase of the final evaluation result and affects the objectivity of the algorithm.In view of the above two points,this paper replaces the indicators in VIKOR with independent prisoner risk characteristics and simplifies the process of weight calculation.There is no importance to distinguish the risk characteristics of prisoners,which avoids the result error caused by index weighting;the characteristics are independent of each other,which reduces the deviation of R value caused by the correlation between indicators.Secondly,this study combines a variety of machine learning algorithms to predict and model the above-mentioned constructed prisoner risk labeling set,and constructs a prisoner evaluation model.The predicted result can represent the risk degree of each risk characteristic of the prisoner.Machine learning algorithms can be modeled based on data training to generate relevant parameters suitable for this data set,avoiding the problem of uniform weighting.Different prisons can generate models suitable for the prison based on the data sets of the prisoners they are in custody,and can also judge the key factors affecting the risk of prisoners according to the parameters.Finally,based on the results of prisoner’s characteristic risk research,this paper designs a risk assessment algorithm based on the characteristics of prisoners’ characteristics based on the interaction between prisoners,and designs a scientific and intuitive prison risk visualization plan.This study quantifies the risk of monitoring and identifies key prisoners in high-risk prisons.It aims to use data visualization to allow prison administrators to visually identify prison risks and help them reduce prison risk through measures such as monitoring and monitoring.The occurrence of the event.