Process Minging And Retrieval Based On Petri Net

Business process mining and retrieval is an important part of business process management.Accurate and efficient business process mining and retrieval can improve the business process management capabilities of enterprises,and thus strengthen the competitiveness of enterprises.With the development of society and enterprises,the logs generated and recorded by the system are more and more complicated,and the process of mining block structures from cumbersome and complicated logs becomes more challenging.By dividing the log vertically,this paper greatly reduces the number of instances of each log partition,shortens the length of each trace,and treats each log partition as the original log.In order to mine accurate models from it,the existing process mining is solved.In the method of discovering the defects of the loop structure and hidden behavior,this paper proposes a behavior block process mining method based on the successor relationship.The method establishes the subsequent relationship matrix of the log according to the successor relationship definition,and analyzes the corresponding values between the transitions in the matrix.Discover all the smallest behavioral blocks and hidden behavioral relationships.Combine all the behavior blocks with the combination principle to get the initial model of the log division,modify the initial model with the hidden behavior relationship to get the final model,and combine all the log-divided models to get the complex log model.In order to improve the efficiency of retrieving models from the enterprise model library and improve the ability of business process management,this paper designs two different similarity measure algorithms from two different perspectives:process similarity algorithm based on transition map edit distance and based on Process similarity algorithm for weighted flow relationships.It is a classic similarity measure to measure the similarity between two models by graph edit distance.This paper gives the concept of transition graph and its generation method,and proposes the concept of length of edge.The cost of deleting and inserting edges is from this edge.The length is determined based on this,the graph editing operation and its cost are defined,and the minimum graph editing distance is calculated by the node matching algorithm.Starting from the structure and behavioral semantics of the model,when measuring similarity,we begin to pay attention to the set of flow relations in the Petri net model.The flow relationship set of a model is often complex,and each flow relationship should not occupy the same component in the similarity calculation.Therefore,the establishment of the weight distribution rule in this paper provides a basis for the reasonable weighting of the flow relationship.Firstly,this paper designs a generation algorithm of weighted flow relationship set.The generation algorithm traverses all nodes and flow relationships except the output library through the breadth-first search method,assigns weights uniquely to each flow relationship,and generates a weighted flow relationship set.Then,The weighted transition set is calculated based on the weighted flow relationship set,and the similarity is calculated according to the similarity definition.There are four main experiments in this paper,which are used to verify accurate models from complex logs,process blocks based on successor discovery,process similarity algorithms based on graph edit distance,and process similarity algorithms based on weighted flow relationships.The experimental results show the feasibility of the four main algorithms.In addition,the similarity properties of two similarity algorithms are tested by the artificially compiled model.The triangular inequality satisfaction rate and running time of the algorithm are recorded.Compared with the existing algorithms,the algorithm is proved to be in some aspects.To some extent better than other algorithms.The main research content and contribution of this paper:(1)An effective method for processing complex logs is proposed.(2)Based on the successor relationship,the concept of the subsequent relationship matrix of the log is given,and the method of finding different patterns is proposed.The method is called the minimum behavior block.This method can find some hidden behavior relations,called implicit direct successor relationship.(3)A process similarity algorithm based on weighted flow relationship is designed.The algorithm traverses all flow relationships through breadth-first search and weights the flow relationship reasonably,so as to achieve different purposes of different flow relationships in similarity calculation.(4)The process similarity algorithm based on the edit distance of the transition graph is aesisned.The algorithm defines the concept of the length of the edge for the first time.By simplifying the model as the transition graph,the node matching algorithm is used to calculate the minimum graph editing distance,and then the similarity between the models is calculated.Figure 36 table 7 reference 41...