The Research Of Business Process Behavior Similarity Analysis Method Based On Petri Nets

As a commonly used method,Business Process Models(BPMs)is widely used to analyze existing business and construct new models in a structured way.The extensive application makes the number of process models more and more large.As a result,the comparison between process models becomes a critical issue to be addressed in many real environments.Therefore,how to quickly find the required model in the process model warehouse has become the focus of attention.These methods are from intelligent warehouse to model similarity search.Among which similarity analysis of models has received extensive attention.The method of behavior similarity measure and evaluation is not only a hot topic of research in the field of business flow but also the basis of many data mining tasks such as block,clustering algorithm and the optimization and reconstruction of some of the business process models.Dozens of documents on the subject is concerned with behavioral similarities,while time factors are weakened.But in practical application,the comparison of models ignoring time information can not reflect the degree of difference among models to a great extent,and has a certain influence on the analysis and research of the model.In this paper,the similarity between the models built by the prototype Petri network is to be analyzed,and then the similarity calculation method SCMBATA by adding time factor in the model is to be put forward.Finally,the clustering method CBOSMA is to be proposed based on the algorithm and the classification results obtained by the clustering method are compared with the traditional SPSS clustering results.It shows the effectiveness of the method.The main contributions of this article include:Firstly,in the trend of the rapid development of computer network technology,a variety of cookies date services based on the Internet have been increased.The users'behavior characters are understood by analyzing and mining cookies network log behavior.The users' network behavior differences can be analyzed by the help of the Petri net model and behavioral relationship theory.Firstly,the relationship between the behavior can be determined by the Petri net model.Then,the existing behavior similarity calculation method based on behavioral profile can be used.The behavioralcharacteristics of two groups is to be compared by used of the similarity calculation method based on behavioral profile and the specific values is to be obtained.Which provide decision support for users'feature analysis.Secondly,in the business process system,behavior similarity analysis has an important research value for the study of relationship between subjects.Which can provide theoretical and methodological support for intelligent recommendation algorithms of social network system.However,the existing research methods are mainly based on the process model structure,and there is a lack of research for business process models with time factors.Based on Sequence Alignment method,Cookies network log can be analyzed by the Delay Petri net.The comparative standard of network user behavior similarity can be proposed,and the fast calculation algorithm of network user behavior similarity can be given.With this method,the similarity of different users can be calculated from the structure and the characteristics of the time Petri net.Which provides a new idea and method for behavior segmentation of network users.Thirdly,Similarity measurement and assessment is a hot topic in Business flow related research areas as well as the core of many data mining tasks,such as segmentation,clustering and so on.Which is the base of the optimization and reconstruction of business process modesl.However,similarity measure in most of the research results have been transformed into the intersection operations of sets,but ignoring other factors of information flow.The factors of a user such as behavior and the length of duration of an action can be considered by use of the Internet users Cookies log.First of all,the user behavior similarity calculation method of SCMBATA can be proposed.On the basis of this,the concept of event membership degree can be used.And then the user behavior clustering algorithm of CBOSMA could be put forward in line with behavior attribute.Finally,using the Web user event log collected by automatic program,SCMBATA and CBOSMA algorithms proposed above can be analyzed and validated.Comparing with the traditional SPSS clustering results,the results show that differences between user behavior similarity with time factors and traditional similarity without time factors is to be obviously different,thus indicating that time will be an important influence factors for network user behavior.