Supporting Structure And Behavior Fusion Process Model Indexing And Retrieval

Process management plays an increasingly important role in the production of enterprises.It enhances the efficiency of different business links with technologies such as process modeling,process mining,and service computing.Process modeling describes the interactions between the work inside the enterprises and clients or partners,which is conducive to the improvement of corporate competitiveness.Currently,large enterprises or organizations have accumulated a lot of process models.This is particularly true in the manufacturing industry where it's commonplace for enterprises to manage tens of thousands of process models.Unlike traditional data,process models contain not only behavioral semantic information,but also the corresponding graphical representations.Therefore,it is a major challenge to retrieve a specific process model.The use of indexes varies according to different inquiry methods.At present,the indexes of process models are all single indexes along a certain dimension.For this purpose,a method is proposed in this thesis for supporting structure and behavior fusion process model indexing and retrieval.The method is specified in the following two aspects:(1)Construction of the process model indexes based on complete finite prefix unfolding: first,complete prefix unfolding is adopted to analyze the process models so as to extract the relationships between activities.Then,contradictory activity relationships are excluded and the relationships of higher precedence are selected for model reconstruction.Parts of the complex structure are made abstract if necessary.Eventually,a process tree is built as indexes after constant iterations of the above steps.(2)Retrieval of process models based on the measurement of process tree similarity: first,the process tree is standardized,based on which,a tree edit distance method is proposed to measure the similarity between the process model indexes.Finally,the similarity obtained by the process tree edit distance method is used to effectively retrieve the process models.The experiment first verifies the accuracy and feasibility of the method for index construction and retrieval.In addition,the performance is analyzed by experimenting with large amounts of data.