Research On Service Discovery Based On Functional Semantics And Process Requiremnts

Oriented Architecture Service (SOA) is a widely used solution to construct and integrate the existing distributed systems. With the development of SOA technology, the number and types of services on internet is growing rapidly which make service resource management more difficult, and how to discover the services for users efficiently and accurately become a significant challenge. In the meanwhile, in order to meet the individual service requirements, the researchers pay more attention to on-demand service model which has been highly valued in the academia. They combine requirement engineer with service computing try to find new service discovery method. Therefore, in the face of such a large variety of service resources, how to achieve on-demand service discovery has become a critical problem.However, existing service discovery methods have some shortcomings: limitation of service document type; lack of an effective way to organize service resource; less paid attention to natural language descriptive services. Besides, the potential progress requirements in user request are ignored in service matching phase. T his paper focuses on the issue that how to improve on-demand service discovery efficiency. The main innovations are as follows:(1) The service requirements modeling method based on functional semantic is proposed to automatically convert natural language requirements description text into business process model BPMN. First, it uses service requirements to extract business activities and process information by utilizing natural language processing tools. And then it using the mapping relationship between workflow pattern and process model to convert requirement into BPMN. The proposed method can not only improve the efficiency of service requirements modeling, but also provide the foundation for service matching based on process requirement.(2) The service clustering based on requirement functional semantic is proposed to reorganize service set. First, it uses service description text to extract the functional information of services by using natural language processing technologies. Based on functional information, obtaining the semantic similarity between two servers and further cluster the services by the K-means algorithm. The proposed method is conducive to quickly locate specific set of service which can meet user’s requirements. This method also provides the foundation for efficient service discovery.(3) The service discovery method based on functional semantics and process requirements is proposed. This method includes two stages of service matching:functional semantic matching and process requirements matching. In first stage, based on service clustering, it locates a specific service cluster class by calculating the semantic similarity and then gets the candidate service set. In second stage, more accurate matching result will filtered by matching potential process requirements and service process model. Experimental results show that the proposed approach can significantly improve the effect of service discovery.In conclusion, in this paper, on the basis of service requirements modeling and service clustering, the service discovery method based on functional semantics and process requirements is proposed. And the experiment demonstrates that the proposed method can effectively promote on-demand service discovery, which has important theoretical significance and application value.