Research On Recommended Method Of Business Process Modeling Combining Disjoint Path And Sequential Pattern

Business process modelling plays a vital role in business process management.In order to improve the quality and efficiency of process modelling,advanced technologies must be used to model business processes.Traditional manual modelling methods are time-consuming and difficult to ensure accuracy.The application of business process recommendation technology can improve the modelling efficiency and accuracy of modelers.Therefore,this paper proposes a recommended method to business process modeling that combines independent path and sequential pattern with the following main work:(1)An edge-disjoint path extraction algorithm is proposed.The representation of business processes is studied,and the Task-based Process Structure Tree(TPST)and the vertex-disjoint path extraction algorithm are analyzed in depth.In view of the problem that the vertex-disjoint path extraction algorithm cannot completely express the TPST’s selection and parallel child node disorder characteristics,and considering the integrity of edge-disjoint paths in expressing the semantics of process behavior,the edge-disjoint path extraction algorithm for business processes is proposed,and the business process model is extracted into edge-disjoint paths.(2)A contiguous path sequential pattern mining algorithm for process recommendation is proposed.In order to solve the problem of low accuracy of business process recommendation method,the advantage of contiguous sequential pattern mining algorithm in sequential data knowledge discovery process is utilized,and arc independent path is taken as sequential data for further sequential pattern mining.According to the characteristics of process recommendation,the contiguous sequential pattern mining algorithm is improved to enhance its effectiveness,accuracy and applicability.(3)A new process node recommendation algorithm is proposed.After pre-processing and mining the process data using the algorithms in(1)(2),a new process node recommendation algorithm is proposed to address the problems of existing methods in terms of business process similarity measures and business process recommendation strategies,using a node recommendation metric to measure the impact of node position on the recommendation results,and using the lift,confidence and support degrees in association rules to measure the Importance.Through experimental evaluation and comparison,it is shown that the proposed algorithm achieves recommendation accuracy of 89.67% and 90.57% on the real and simulated datasets respectively,while ensuring time efficiency.(4)Development of a visual prototype system.Based on the proposed approach,a visual prototyping system has been developed that includes functions such as recommend,add,delete,edit and drag-and-drop nodes.When a user selects a recommended node displayed in the system as a hit point,the system automatically builds the next process node based on the user’s selection,without the need for manual construction.Test experiments have confirmed that the proposed system can effectively support practical applications.This paper shows that extracting the edge-disjoint paths of a process through the edge-disjoint path extraction algorithm and further mining the behavioral knowledge of the process using the contiguous path sequential pattern mining algorithm to recommend process nodes by combining node recommendation,process matching,confidence,lift and support can effectively facilitate efficient and accurate modelling.