Research On Service Matching Technology In The Process Of Transformation From ARTIFLOW To BPEL

Artifact-centric business process management is a new tendency of the development of BPM. In this trend, a new Artifact-centric process model- ArtiFlow came into being. The emergence of this model integrates BPM and SOA as a whole. It records the key data of process using Artifacts and completes the lifecycle of process according to the management from service to Artifacts. It combines data and process into one. Transformation from ArtiFlow to BPEL became a key issue of Artifact-centric business process management, and the matching between logical services and physical services is an important part of transformation from ArtiFlow to BPEL. After analyzed the research status home and abroad comprehensively, this paper have done deeply research about service matching on the transformation from ArtiFlow to BPEL.Firstly, on the basis of analyzing the relationship between services and Artifacts in ArtiFlow logical models, the services can be divided into three categories according to the effects of services on Artifacts, and the roles of services in each category are described following.Secondly, this paper deeply analyzed and studied the ArtiFlow logical models and technologies of service matching, so that to solve the problem of service matching in the process of transformation from ArtiFlow to BPEL. On the basis of that, a new semantic service matching method between different layers which fits for the transformation from ArtiFlow to BPEL was proposed. It introduced three types of interfaces dependency of services in the process of matching so that to make a high matching recall and precision.Thirdly, in order to find the services fits for business process from large number of physical services more efficiently and accurately, we introduced operation tree model and use it to describe the functional capabilities of services. In this paper, the relevance between services is also considered to ensure that the services obtained can complete the whole business process successfully.Finally, the feasibility and effectiveness of the algorithms are validated by experiments, and the experiment results are analyzed.