Through machine learning on more than 130,000 legal documents since 2020,this study re-veals that when making detention decisions,Chinese judicial personnels assess the social dangerousness of criminal suspects,but such assessments focus on elements of"criminal dangerousness"while overloo-king those of"personal dangerousness".The introduction of the quantitative assessment method is main-ly beneficial for the handling of"complex"cases in which it is difficult to accurately determine whether to detain the suspect relying solely on basic case informations.It is argued that the quantitative method has the potential to significantly reduce the detention rate in such cases.In the current practice of deten-tion decisions,case handlers mainly rely on subjective assessment based on the elements of"criminal dangerousness".Traditional reform path focuses on rule refinement and element reconstruction,but falls short of addressing the insufficient assessment of social dangerousness.Incorporating big data modeling methods can help to systematically integrate scattered detention decision information.In addition,we can also expand the training data of the model by using methods such as questionnaires,scales,and dig-ital devices,and further improve the accuracy of quantitative tools by deepening the understanding of the mechanisms underlying social dangerousness and distinguishing between causes and types of danger to construct multiple sub-models.
arrestsocial dangerousnessquantitative assessmentnecessity review of detention
NETL
NSTL
万方数据
