Research On Key Technologies Of Multiple QoS Constraints For Web Services Discovery

With the drastic development of Web technology, modern software systems are towards the tendency of decentralization, platform independence and loose coupling. Software systems need to change themselves to adapt newly changing requirements. The emergence of Web services residing in Service Oriented Architecture (SOA) satisfies such a tendency. As relevantly independent software function entities, Web services have been used as the next generation paradigm for application integration, and have been universally applied in E-commerce, E-government, and enterprise application integration, etc. In recent years, the rapid development of scientific computation has boosted the great progress for Web service based management of workflow systems. Different from traditional control flow based ideas, recent Web service based management of workflow systems have paid more attention on data flow information, which concentrates on the data transmission between Web service nodes. Therefore, it makes data oriented workflow systems more suitable for large scale of scientific computation and big data tasks analysis such as genomic computation, earth science, astronomy computation and so on.However, users information and their specific requirements for workflows have not been fully considered in the retrieval research of data oriented workflows. Workflows retrieval in the past is based on programmers or experts of computers which requires some computer theory and technics. This may be difficult for these true users who only care how to use these systems to solve their specific requirements. This situation is also a challenge for efficiently reusing of data oriented workflows. Recent workflow retrieval does not pay much attention on semantic information deep in each task/service node. This may cause the difficulty on recognizing these workflows with same function but different names because they come from different service providers. Due to the open, heterogeneous platform that data oriented workflows run on, semantic based data oriented workflows retrieval seems to be a more effective method which also considers the users demand of quality of services such as responding speed, service cost, energy cost, and so on.This paper will study the processes discovery technics of data oriented Web services. First, a formal representation based on graph is proposed to model semantic based on data oriented workflows with variable time constraints. Ontology languages and data types are used to represent the semantic information of task nodes and the input and output parameters. It is the theory basis for further analysis. Second, a framework is introduced for workflows retrieval with variable quality of services. It seamlessly combines the semantic information and variable quality of services. Third, we discover the specific workflows retrieval method based on this formal representation and give a distance measure to compute the similarity of data oriented workflows. All candidate workflows will be ranked based on the similarity measure and users will find it easy to choose a suitable one to finish their tasks. Finally, we made the related experiment to validate and evaluate the effectiveness and efficiency of our method.This research will provide a novel idea and way of studying the data oriented workflow retrieval. It will be helpful to improve the level of automation operation for huge data computation. Important significance will be shown in practice and academy especially for the development of recent scientific workflows research, scientific computation and big data analysis.