摘要：

大数据促进了司法证明方式的革新。相较于传统证明，大数据证明具有超越人类主观经验的智能化特征，并从物理空间转向数据空间，从侧重逻辑的因果关系转向侧重数理的因果关系，从面向过去的证明转向涵摄未来的证明。数据、算法及法律程序是影响大数据证明可靠性的主要因素，具体表现为数据法律层面的错误及样本不全面，算法模型不准确及不公正，法律程序不透明。对此，可构建基于数据规则的可靠性审查机制。数据层面，应加强法律数据真实性判断，保证数据的全样本。算法准确性层面，引入算法同行评议机制，保障算法适用情景的匹配性；算法公正性层面，选择多元化数据集、识别替代性变量以及审查模型偏见。法律程序层面，建立算法开示程序、大数据证明结果排除规则以及专家辅助人出庭等制度。

关键词:  

Abstract: Big data has promoted the innovation of judicial proof.Compared to the traditional proof，big data proof has proved to be intelligent beyond human subjective experience，with the features from physical space to data space，from causality based on logic to causality based on mathematical theory，and from proof oriented to the past fact to proof involving the future.Data，algorithms and judicial procedures are the main factors affecting the reliability of big data proof，which are reflected in the inauthenticity of data and incomplete data samples，inaccurate and unfair algorithm models，and the opacity of legal procedures.Therefore，a reliability mechanism based on data rules can be constructed.In terms of the data，it is suggested to review the authenticity of data in law to ensure the complete samples.In terms of the accuracy of algorithms，it is suggested to introduce the peer-review mechanism and focus on the adaption of the algorithms’ application scenarios.In terms of the fairness of algorithms，it is reasonable to prefer the diversified training data set，identify the proxy variables and inspect the model biases.In terms of the judicial procedure，it is suggested to establish the rules of algorithms discovery，the exclusion rules of big data proofing result and the auxiliary experts testifying in court.

Key words: Big Data, Judicial Proof, Algorithm Bias, Algorithm Disclosure