摘要
An intelligent methodology and its prototype system are developed for automatically discovering legal precedents using semantic analysis. The concept of the trademark legal precedent recommendation was originated from our TE2020 conference paper. The approach is to identify matching cases related to given seed case with respect to their legal case brief attributes using advanced text mining techniques. In the paper, dynamic topic modeling is further developed to analyze the dataset over three time-sequential cohorts to identify trademark law topics varied over time. Further, the prototype system was demonstrated and verified using real trademark case analysis with satisfactory results.