Algorithm Research Of Conformance Checking Of Process Mining

Process mining aims at discovering process models from workflow log with mining algorithm,that is,extracting information related to the activities and relationship between activities from event logs of some of the information system, capturing the business process as the enterprise information system is being executed. Conformance checking, also referred to as conformance analysis, aims at the detection of inconsistencies between a process model and its corresponding execution log, and their quantification by the formation of metrics. The process of the conformance checking is that we can check whether the process in the real condition is conform the pre-defined process model based on running event logs, and also we can check the deviation and the consequence between them. Many researchers have proposed an incremental approach to check the conformance of a process model and an event log. First of all, the fitness between the log and the model is measured(the observed process comply with the control flow specified by the process model). Second, the appropriateness of the model behavior and model structure can be analyzed with respect to the log(the model describe the observed process in a suitable way). We can quantitatively calculate the conformance between actual production activities and process model through conformance checking techniques.Moreover, the application of the techniques can help to compare the advantages and disadvantages of model structure, provide theoretical supporting for the standardization of modeling and check the conformance between the process model in the management systems and the real business process to improve the process model efficiency and the production services. Therefore, it is of great significance and it is currently one of the hot research topic in the field of process mining.Although classical conformance checking techniques can identify deviations of process executions from predefined models,but there are still some defects:1. fitness,conformance checker statistics for more data,unable to support complex log, and if all events of a log have no relative transitions,fitness metric has no significance.2.behavior precision, conformance checker converts the model into the minimum coverage map,results in high algorithm complexity and is restricted by the problem size, has some limitations.3.structure precision, only considers duplicate task and redundant hide task the two kinds of nodes,without considering the model may contain redundant places. When the fitness,behavior precision and structure precision are of the same value between the two models with the same log, unable to make further distinguish, also cannot help people to determine the best model.This article improves checker method.The main contributions include:1.Improve fitness calculation, so that it can handle more cases and be more effective in dealing with the complex log content.2.The paper integrated the core idea related to conformance checker and come up with a simple new method to caculate behavior precision.3.Put forward further judgment standard for structure precision to select the optimal model if the first three measurements are consistent.The conformance checker+ idea presented in this paper has been implemented within the Pro M framework,after experimental verification, checker+ achieves improvement goals for check.