关键词: 业务流程模型抽象,聚类分析,虚拟文档,行为聚合

Abstract: ion Based on Cluster Analysis SUN Shan-wu^1,2 WANG Nan^1,2 OUYANG Dan-tong³ (College of Management Science and Information Engineering,Jilin University of Finance and Economics,Changchun 130117,China)¹ (Laboratory of Logistics Industry Economy and Intelligent Logistics,Jilin University of Finance and Economics,Changchun 130117,China)² (Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education,Jilin University,Changchun 130012,China)³ Abstract A prominent business process model abstraction(BPMA) use case is a construction of a process “quick view” for rapidly comprehending a complex process.Some researchers propose the process abstraction methods to aggregate the activities on the basis of their semantic similarity.Most of these methods focus on k-means cluster analysis which partitions an activity set into k clusters(k is predefined by users) and assigns an activity to a cluster whose centroid is the closest to this activity.But in fact,the number of the abstract activities(clusters or subprocesses) is unknown so that which activities are assigned to the same subprocess tends to depend on the designers’ experiences or modeling habits.Moreover,if the activities are not part of particular cohesive groups from the point of view of experienced mo-delers,the initial merging decisions may contain some errors.In this paper,the virtual document was introduced to represent activities and process models in order to eliminate the constraint of using the particular attributes of activities as the dimensions of vector spaces.And an approach was presented that exploits an industrial process model repository to derive a distance threshold between an activity and the subprocess it belongs to.According to this threshold,an algorithm was proposed to generate a possible number k of the clusters.By using k as a parameter and the distance threshold as a further constraint,a clustering procedure was given to aggregate the activities into different clusters.In an experimental validation,we compared both the proposed approach and k-means clustering(without the distance threshold as a further constraint) with actual model decisions,showing a stronger correlation between the proposed approach and actualmodel decisions.As such,this paper contributes to the development of modeling support for effective process model abstraction,easing the use of business process models in practice.

Key words: Business process model abstraction,Clustering analysis,Virtual document,Activity aggregation