Development of a process mining system capable of handling temporal and compositional irregularities

在这项研究中，我们开发了一种基于典型性指数的常见处理过程组的发现方法，该指数反映了下达命令的全局出现和转换频率。我们对158例耳鼻咽喉疾病数据进行了实验，观察到应用路径的分布在所生成的四个聚类之间具有统计学意义上的差异，以评估排序序列的典型性指数的相似性。同样，在产科疾病数据(124例)上，我们观察到应用路径和生成的簇的分布之间存在统计上的显著差异。基于这些结果，我们证明了通过使用典型性度量，我们可以发现相似处理过程的组，而不需要给出关于订单类型的任何先验信息，例如操作。

In this research we developed a method for finding groups of common treatment processes based on the typicalness index, which reflects the global occurrence and transition frequencies of issued orders. We conducted experiments on otolaryngological disease data (158 cases) and observed statistically significant differences on the distributions of applied pathways, among the four clusters generated assessing the similarity of the typicalness indices of order sequences. Similarly, on obstetric disease data (124 cases), we observed statistically significant differences between the distribution of applied pathways and generated clusters. Based on these results, we demonstrated that by using the typicalness measure we could discover the groups of similar treatment processes without giving any a priori information about order types such as operations.