Process Model Repair Techniques For Special Choice Structures

Process mining is an important tool used by modern organizations to manage complex operational processes.Process mining aims to extract valuable process-related information from the event logs of enterprise information systems and build process models.Establish effective associations between process models and event logs generated by modern information systems.There are three types of application scenarios for process mining:process discovery,conformance checking,and process enhancement.The process model can describe the characteristics of the system,verify the correctness of the system,and provide feedback to the system designer.However,there may be some deviations between activities in the event log and the process model due to system upgrades or business process improvements.The process enhancement application scenario of process mining can solve the problem of deviation.The process of correcting deviation is called model repair.Existing model repair methods focus on the quality dimension of the fitness degree,and give less consideration to other quality dimensions.The repaired model contains repeated sub-processes or self-loops,and the model has redundant parts.Although the repaired model can replay the event log,the model is very different from the actual business process and cannot accurately reflect changes in the actual business process.For this reason,for different model structures,this paper proposes different methods of process model repair based on logic Petri nets.(1)For the model containing the choice with concurrency structures,a logic-based Petri net is proposed to repair the process model with the choice with concurrency structures.First,in order to find the position of the deviation in the model,the concept of the simplest alignment is proposed,and the reachability identifier of the Petri net is introduced to help determine the location of the deviation through the reachability identifier set.Then,the logical relationship between transitions is excavated at the position where the deviation occurs,and a logical function is added to control the trigger of the transition,and logic Petri net is constructed.The repaired model can accurately reflect the actual business process.Due to the addition of logical relationships,the relationships between activities are more accurately expressed using logic Petri net than classic Petri net.(2)For models containing non-free-choice structures,a method for repairing process models with non-free-choice structures based on logic Petri nets is proposed.First,the concepts of transition pairs,post-order relations and post-order transition matrix were proposed to help identify non-free-choice structures.Then obtain the transition pair from the event log,and then obtain the post-order transition matrix according to the post-order relationship in the transition pair.In the optimal alignment,the transition pairs are traversed to obtain the abnormal transition pairs in the post-order transition matrix.Through these abnormal transition pairs,the position to be repaired can be determined,and finally the logical relationship between the transitions can be mined,and the original model can be repaired based on the logic Petri net.Due to the added logical relationship,the relationship between activities can be described concisely,which makes the model more concise and avoids complicated and redundant structures.Finally,through simulation experiments,the correctness and effectiveness of the two special structure repair methods in this paper are exemplified,and the obtained model has more ideal results in terms of precision and simplicity.