Online Process Model Generation And Evolution Method For Service Management

As an important means of log analysis in enterprise information systems,process mining is widely used to understand the runtime process execution logs,to discover process models from multiple perspectives and to combine service management with mining results.Nowadays,under the background of emerging Internet applications such as mobile APPs,some problems related to log data preprocessing,process model generation and comprehensive result analysis still exist in traditional process mining techniques.In detail,first of all,compared with the traditional process-aware information systems,the original logs of most mobile APPs lack the case description for distinguishing different process instances.Secondly,previous studies which focus on single-version process model generation based on static logs make the discovery slowly sense the changes of the model and lack the continuous analysis.Thirdly,the service management which is integrated with mining results emphasizes the feedbacks of the single version but lacks the consideration of changing process and the comparison with multiple generated results from the time perspective.To solve the above problems,an online process model generation and evolution method for service management is proposed in this paper.Aiming at analyzing the changing business process,our method firstly converts the original service log stream to standard event stream based on field mapping and deduction.Then the continuous process discovery and model versioning are carried out.Finally,based on the generated multi-version process models,the evolutionary relationship is analyzed,laying the foundation for service governance.The main contents of this paper are listed as follows.(1)An online process model generation and evolution framework for service management is proposed.An online process model generation and evolution framework for service management is proposed to achieve the continuous analysis of changing process for both process discovery and evolution analysis on models.Under the consideration of the changing process and continuous disposing,the log collection and generation layer,the online mining layer and the monitoring layer are established for conversion from service logs to event logs,continuous process model discovery and service governance with model evolution analysis.(2)A conversion method from the service log stream to the event stream is implemented.A conversion method from the service log stream to the event stream is fulfilled.Operation context is taken full use of in deducing missing Case ID.The implementation scheme for the conversion method in a streaming environment is also discussed,generating the standard event stream for next steps.(3)A continuous process discovery method from the event stream is introduced.A continuous process discovery method from the event stream is implemented to generate process models and to perform model versioning.Integrated with several methods on concept drift detection and stream process discovery,the developer-driven,as well as user-driven changes can be caught,resulting in multi-version process models.(4)A method to apply evolutionary relationship analysis among multi-version process models to continuous service governance is discussed.Based on continuous mining results,this paper focuses on the evolutionary relationship calculation based on footprint matrixes.Through the analyzed results,executed services are classified and the classification results are applied to continuous service governance.(5)A process mining based service governance platform is constructed.According to the proposed framework,a process mining based service governance platform is constructed for mobile APPs.Taking an m-health APP as a background,the implementation and application scenarios of our methods are introduced.A series of experiments as well as case studies are performed to illustrate the effectiveness of our methods.