Study On Process Variants Clustering Mining Method Based On Behavioral Profiles

In the past few years,a new generation of process aware information system has emerged.It can configure process models at build time and change process instances at run time.Due to the change of application and the adaptability of process model,there are a lot of process variants from the same model in real application.Many similarities in structure and function are existed among these process variants,but at least some details and structures are distinguished from each other.It is difficult to configure and maintain these variants.At present,the research on process variants is mainly focused on variants mining.How to mine the adaptive original model from a variant set is the primary goal,and then how to better distinguish the differences between variants is also an interesting aspect.In order to solve these two problems,this paper adds the behavior relationship between activities to the research of variants based on behavioral profiles.It enriches the content of process variant research,improves the accuracy of the methods used,and expands the application prospectsFirst,for the process model and its variants,the difference between variants cannot be measured with a single distance when the distances are the same,.In order to solve this problem,the concept of behavioral profile consistency is introduced.Distance and behavioral profile consistency are combined to evaluate the correlation between process variants and original reference model quantitatively by behavioral profiles consistency distance.In this way,6 process variants with equal distance can be distinguished by profile consistency distance.The existing research of conclusion on the relationship between variants and original process is enriched.In order to use less system model to describe process variation,a process variants clustering mining algorithm is proposed.Firstly,the variant model is matrix combined with the concept of causal behavioral profiles.Secondly,the weighted cosine similarity is applied to cluster activities and determine the behavior relationship between them.Finally,through repeated iterations until clustering all activities,a target reference process model is obtained.It is easier to configure different process variants than the original reference model.The simulation results show that the algorithm is feasible and effective in mining reference model.Next,aiming at the problems of different activities and circulation in variant set,the algorithm of process variants clustering mining is improved and the practicality of the algorithm is verified in the actual case of auto insurance claims..Based on the behavioral profiles,the consistency distance and two related mining algorithms of process variants are discussed in this paper,the main contents are as follows:(1)The concept of consistency distance based on behavioral profiles is proposed to distinguish process variants quantitatively,especially when the distances between them are equal.(2)The process variants mining algorithm based on causal behavioral profiles is proposed.It is used to mine and process simple variation sets without cyclic structure,so as to obtain an adaptive original model.Experiments show that this algorithm is more convenient and efficient than the traditional process mining algorithm and the classical heuristic algorithm.(3)The improved algorithm for mining process variants is proposed.It is mainly used to solve the two problems of different types of activities and cycle modules in process variation,so that the algorithm is more perfect and the application prospect is more extensive.(4)The practicality of the algorithm is verified by the actual case of auto insurance claims.Figure 11 table 13 reference 65...