Research Of Business Process Decomposition And Mining Method Based On Event Log

In the age of information,the rapid development of the Internet has greatly changed people's lives.Most of the information are recorded on the Internet and corresponding business process information is generated.For enterprises,the analysis of business processes can obtain the execution of daily behavior information and the background of abnormal occurrence,so they can adjust and optimize the internal working structure of themselves.Process mining technology can use log information to achieve this goal.By processing the event log,the information in the log can be shown as completely as possible in the mined process model.However,with the continuous development of big data,the data received by people is growing explosively,and a large amount of user-related information is also recorded inside the enterprise.Process mining techniques have scalability issues when applied to large event logs,both in terms of the computational requirements and the interpretability of the generated output.Because process models discovered from large event logs often provide limited insight and the resulting models often have problems of low precision and low adaptability.In order to solve problems above,the common solution is to decompose the process into various stages,and carry out separate process mining for each stage and establish a separate business model.Finally,the sub-models are merged to form the final business process model.Based on the background above,this paper studies the method of process decomposition and mining based on event log.In order to solve the problem of the low precision of process mining,this paper proposes a graph-based clustering process decomposition technique.This method selects candidate segmentation nodes based on the maximum flow algorithm,and performs module clustering by maximizing the modularity difference.The result of decomposition can achieve an effect that closer to manual decomposition.In order to solve the problem of high complexity and low accuracy of the mining model,the article gives a process mining method based on sub-process set.First,construct the set of sub-processes into time-based concurrent flow graph.Secondly,aggregate and simplify the sub-process models mined in order.Finally,complete the details of the process model based on the business process information to form the final business process model.In the experimental verification part,this paper analyzes both the process decomposition and decomposition-based process mining in detail.This paper verifies the graph-based clustering method proposed in this paper by comparing the mining algorithm that generates a flat model and the mining algorithm based on decomposition.The effectiveness of decomposition algorithms based on graph clustering and process mining methods based on decomposition.