Researches On Mining Methods Of Business Process Based On Behavioral Profiles Of Petri Net

Business process mining is a new kind of data mining applications in the field of business process management. In order to achieve the best mode of the business processes, the data of logs recorded in information of business processes are analyzed. In order to make the business processes revert to the real process models, and use the log sequence to analyze and optimize the business processes. It is conducive to adjust and optimize the structural state of the original business process models, reduce operating costs, improve operational efficiency of software, and create greater value for the enterprises.Currently, the researches for process mining methods are largely for the workflow model mining; due to the development and advancement of business process management systems, the requirements for business processes are continuously improved. Compared with artificial business process modeling, the subjective factors are reduced in process model mining from the log information, making the model more objectivity; other words, the mined model not only meets the required business functionality, but also has constraints on behavior in the model, reduces redundancy of models to make the model more concise. Conventional mining methods mostly rely on direct dependencies between activities of the model. The paper add behavior profiles into the mining process, and use the consistency between logs and models and the consistency between models based on behavior profiles to optimize the mining model. The main contributions of this paper are followed.(1) An optimized method of business process mining based on the execution logs is proposed. Business process mining is a method to improve the operational efficiency and the quality of service, which has become a hot research. As more data of log files become available, the mining technology of process has also been a large degree of development. Process mining technology is mainly divided into two parts:one is to obtain models from the log data, and the second is the consistency test. Most of the existing methods for business process mining are based on the analysis of the structural properties of Petri nets, thus there are limitations in the construction of a behavioral model. A mining method of business processes based on behavioral profiles of Petri nets is proposed. First, the behavioral order relations are constructed; then present the measurement methods of consistency analysis of models based on behavioral profile, and on this basis, the mining method of business processes is proposed. Finally, simulation analysis is conducted on the PROM5.2platform, and the results show the advantages of the method.(2)The mining method of business processes based on L*algorithm and incremental logs is proposed. With the continuous development of business process systems, the number of logs is increasing recorded in the information system; in order to avoid excessive number of logs in one-time extraction and thus result in excessive computing, this method uses incremental logs for business process mining. Adopt the idea of query in L*algorithm to find the relations of behavior profiles of logs. Build initial models according to the basic structure of behavior profiles of logs based on Petri nets; use the incremental logs to optimize models; in the optimization process, use the contour the consistency based on behavior profiles to test. Finally, a simple example is used to verify the feasibility of the method.