Log automaton under conditions of infrequent behaviour mining

In the existing process mining methods, infrequent behaviours are often considered as noise is ignored, but some infrequent behaviours have an important role in business process management. Firstly, the knowledge of log automaton is applied to the low-frequency log to delete infrequent behaviour in the logs; secondly, the processed logs are added into attributes. Then, the condition-dependent value of the communication characteristics of different module networks is compared with the threshold, and the effective infrequent log is retained to optimise the model. Finally, a practical case is applied, which indicates the effectiveness and validation of the proposed method.