Research On Alignment Calculation Method Based On Process Structure

In process mining,conformance checking of business process models quantifies the deviations that occur in their actual execution.While detecting where deviations occur,and the set of deviations,is a very important issue,conformance checking has more far-reaching implications for organisations and many other process mining techniques.For example,it provides a new way of enhancing processes in model repair techniques.Essentially identifying these deviations can be attributed to conceptual alignment.Conformance quantifies the extent to which a process model can provide an overview of its observed behaviour,i.e.the event log.In fact,an alignment is a matrix data structure with only two rows,where each column indicates whether an observed event can be aligned with the corresponding node of a given process model.That is,the optimal alignment is the best combination of corresponding behaviours that the process model can mimic as observed.Current alignment calculation techniques only consider the results of optimal alignment in terms of the number of deviations that occur,and their exponential time and space complexity hinders their applicability in medium to large sized cases.The main objective of the paper is to propose an enhanced activity-dependent optimal alignment calculation method based on a process model structure and to reduce the time and space complexity of the alignment calculation based on reduction of the model structure.Beyond this the goal is to address the above challenges from the perspective of the overall situation,i.e.without considering all details.In this case,the focus of the research will be on extremely important labels or positions.Doing this not only alleviates the computational challenge,but also provides a greater logical perspective of deviations.The main focus of the article is to give a Petri net model optimisation case,an explanation of Petri net related definitions and a demonstration of alignment calculation methods on top of this,followed by a proposed label weighting method based on the structure of the process model.Then,a computational space compression method based on model reduction is proposed based on graph theory,and how to extend it with other computational methods is demonstrated.For model optimisation,activity label dependence in optimal alignment,and algorithmic time and space complexity,this paper contributes the following research based on the theory and applications related to Petri nets.(1)In recent years,new transaction models such as self-checkout and unmanned payment in supermarkets have emerged.Here the article uses Petri nets to model the supermarket self-checkout system and analyses the system’s activity,boundedness and other related properties using HIPS;it combines the theory related to behavioural contours to optimise the net system,and also uses CPNTools to analyse the optimised system in terms of time delay waiting.The optimised net system reduces the overall payment process elapsed time and reduces the contact time between people to a certain extent.(2)In response to the existing alignment calculation process in which different activity labels appear and existing alignment calculation methods use the same cost for different activities,a label weighting method based on the model structure is proposed here.The method analyses the direct following,selection and concurrent structures contained in the model and calculates the weights of different activities while normalising them.The method is able to differentiate the different activity weights in the calculation of the optimal alignment and reflect the results of the model in calculation(1).(3)To address the problem of time and space complexity in the process of optimal alignment calculation,an optimal alignment calculation method based on model reduction is proposed here in conjunction with some of the methods in graph theory.Here the model and observation behaviour is reduced without losing alignment information,and the structural simplification not only reduces the search space for the alignment calculation,but also provides an overview diagram of the alignment calculation.On top of this,a partitioning strategy will be provided for the method in(2),decomposing the original problem into a smaller set of solvable independent problems.Figure [28] Table [8] Reference [70]...