Conformance Checking Of Model In The Environment Of Process Mining

In process mining,after we get the model from the log by using mining algorithm,the coexistence of process model with event log brings such a problem: whether the model and the event log are mutually consistent.The conformance checking means that by comparing the model getting by the model discovery with the actual process’action or the event log,we can find that whether the model is factual.We can also use some metrics to show the level of conformance.Fitness shows the the scope of the contact between the event traces in the logs and the effective execution path in the process models.It also means whether the the event traces in the logs can map to the effective execution path in the process models.The measure of the fitness is a very important step in the process mining.Log is made up of many records.It means the every record is a actual reflection of the running of the system,although this reflection can’t be reflected by the model.In other words,every record can affect the fitness. So we can analysis every record and use the statistical method to measure the fitness.Based on the above theory,this article gives a new algorithm to measure the fitness by using statistical method.Every Fitness between each record and the model can be used to measure the Fitness between the whole log and the model.By analyzing the existing methods to measuring the Fitness,we can get a method to measuring the Fitness between the single record and the model from the method named Token Replay.All the Fitnesses can be regarded as a sample space.And then we can used the statistical method to get the fitness between the whole log and the model.Based on this,we put forward the concept of Stability.Because the whole Fitness is the reflex of the every Fitness between the each record and the model.By measuring the deviation of the each Fitness and the whole Fitness,we can get the log’s status of development and degree of fluctuation.Stability is used to quantitative describe.Through the calculation of Stability, we can analyze the Stability of the log and to estimate the development of the log in the futureExperiments show that this algorithm can measure the fitness accurately.The Fitness is more bigger, the marching degree of the log and the model is more better.The Stability is more bigger,the log is more stable.