| With the rapid development of information and intelligent society,the role of business process management is becoming more and more prominent.A business process model with superior performance can significantly improve the work efficiency of the company system and increase its core competitiveness.As a main technical means in business process management,process mining includes process design and implementation,model establishment and optimization.How to efficiently use process mining technology for log data modeling and configuration optimization is of great significance.In the actual process of mining process model,there are often some deviations or exceptions between the event log and the mining model due to some factors.Process mining technology needs to standardize the executable business process.The existing technology has some shortcomings in explaining and emphasizing whether the process model can be adjusted.Therefore,this paper proposes a method to identify configurable elements in process model and add configuration changes to workflow model.Configurable elements are the transitions of workflow models that can be modified,so that the behaviors represented by the models are limited.A configurable element can be set to allow,block,or hide.For hidden transition mining,our method is to identify the hidden transition through sequence coding graph,log automata,block structured Petri net and binary search tree to mine the ideal model with configuration elements.The main contributions of this paper are as follows:1.Based on business process event log,a method of mining business process model is proposed.By analyzing the behavior profile relationship of log sequence,the optimization problem of the model is found.At the same time,the change optimization model is configured from the perspective of finding change domain and function block,which is verified by the book borrowing system in university library.2.Based on Petri net block structure,a method of mining hidden changes in process model is proposed.For a given complete log set,before process mining,the event log is preprocessed,and the sequence coding graph of the event log is drawn,and the frequent sequence and the non frequent sequence are distinguished.The algorithm is used to mine the initial model for the frequent sequence,and then the behavior contour block structure is used to decompose the initial model hierarchically to get the block structured workflow network.By matching the nontrivial subsequence with the decomposed model block structure,the possible hidden transitions in the block structure are identified.Finally,from the two evaluation indexes of fitting degree and accuracy,the suspected hidden changes are screened out,and the abnormal changes are filtered out,so as to mine the target model with configuration information.Through a specific example,the feasibility of this method is verified.3.Based on binary search tree,a process model mining method is proposed.When the log set is complete,the hidden changes in the log are mined by depth first search(DFS)of binary search tree.Firstly,after the initial cleaning of the log set,the activity relation table and the log automata of the log are made in turn,and the relative frequency of the arc is used to distinguish the frequent log and the low-frequency log.The updated log automata are split to mine the block structured workflow net,and then the low-frequency traces are replayed in the initial model,and the minimum editing distance based on different cost functions is calculated The effective sub segment is analyzed and extracted,the effective sub segment is searched in depth first on the binary search tree,the change range is located,the hidden transition is configured to get the target model,and the consistency evaluation index is used to evaluate the model,and finally a more perfect and ideal target model is obtained.The feasibility of this method is verified by a process example.Figure[18]Table[7]Reference[85]... |
PDF Full Text Request