Abstract: Business process event logs sometimes contain chaotic activities,which are a kind of activity independent of process state and free from process constraints,and may happen anytime and anywhere.The existence of chaotic activities can seriously affect the quality of business process mining,so filtering chaotic activities becomes one of the key contents of business process management.At present,the filtering method of chaotic activity mainly filters infrequent behavior from the event the log,and the filtering method based on high frequency priority is not effective in filtering chaotic activities in the log.In order to solve the above problems,a method based on log automata and entropy is proposed to filter chaotic activities in logs.Firstly,a suspicious chaotic activity set with high entropy is obtained by calculating the direct preset rate and direct posterior set rate of activity.Then,the log automata is constructed from the event log.From the log automata model,the intersection of the activity set of infrequent arc and the activity set of high entropy in the log is calculated to obtain the chaotic activity set.Finally,the conditional occurrence probability and behavior profile are used to determine the dependence between the chaotic activity and other activities,so as to decide whether to delete the chaotic activity completely in the log or to keep the chaotic activity in the correct position in the log to delete other activities.The effectiveness of the method is verified by case analysis.

Key words: Behavioral profile, Chaotic activity, Conditional occurrence probability, Entropy, Log automaton, Petir net