An associated probability analysis model based on multi-element fusion

In judicial adjudications, the chain of custody (or chain of evidence) is the core element of conviction. A complete chain of custody can provide a solid basis for judicial trials and sentencing. However, the evidence correlation analysis in the litigation process mainly depends on the judge's manual judgment, which often has the problems of poor interpretability and strong subjectivity. At present, the use of data mining technology to analyze the judicial data is of great significance and practical value in the construction of smart courts. This paper focuses on the problem of multi-factor modeling and reasoning about the trusted relationship between elements, and proposes an association probability analysis model based on multi-element fusion. Also, the process of obtaining the chain of custody of the judicial field is used to further illustrate the specific construction process of the model. Firstly, the factual decision chain is defined based on the knowledge of the judicial domain, and a multi-evidence association network is constructed, in which each evidence entity acts as a node in the network, and calculates the correlation probability between nodes through the association relationship between the evidence elements. Then, according to the event decision chain, the association constraint relationship between the evidences is established through the Bayesian network. Therefore, a multi-evidence association model is established, and all possible chain of custody combinations are obtained through genetic algorithm, so as to optimize the multi-evidence correlation model. Finally, the optimal chain of custody is obtained through the model. The validity of the model is verified by experiments in three types of cases on real data sets.