Establish Evidence Chain Model on Chinese Criminal Judgment Documents Using Text Similarity Measure

One of the most prominent issues in criminal judgment documents is the insufficient evidence. In order to raise the level of judgment documents reasoning, we need to evaluate the quality of evidence in judgment documents. In the recent informatization of Chinese courts, the huge amount of law cases made it necessary to automate the evaluation of evidence. Constructing the model of evidence chain is the basis for assessing the quality of the judgment documents as evidence chain model can describe the relationship between evidence and fact as well as the relationship between evidence more intuitively. In trying to achieve all above mentioned, we propose a model of evidence chain based on Chinese criminal judgment documents. Automated text preprocessing of Chinese criminal documents creates semi-structured XML documents and in XML file, we can get evidence set and fact set. Key element extraction based on syntactic parsing is used to get keywords of each evidence and fact. Text similarity measure based on Word2vec and keyword overlap ratio calculation is used to get the connection point of evidence chain. Predefined weight of different kinds of evidence can be used to measure the importance of each evidence chain. Table format and graphical format make it possible for us to see the structure of evidence chain.