Research On Method And Key Technologies Of Semantic Model-driven Development For Services Oriented Enterprise Application Software

With the rapid improvement of information technology, Internet has been playing an important role in developing enterprise application software. The enterprise application is no longer a closed system in that new properties like heterogeneous, cooperative, and dynamic are introduced. Both the software form and developing model are changing rapidly. Service oriented software development has become a new paradigm. Software service is becoming the standard form of resource encapsulation(?)ring in the open network environment. It is the basic function unit of application system. In the future, the enterprise application tends to concentrate on business process, and be developed by reusable services discovery and dynamic composition. In this background, the dissertation focuses on the studies of enterprise application development. The major work and contributions are as follows:First, propose a semantic model-driven development method for services oriented enterprise application software:SMDA.In this method, the Service-Oriented Architecture (SOA) serves as the basis architecture, Model-Driven Architecture (MDA) is the basic development paradigm, and the ontology technology provides semantic support. The three technological spaces are combined with UML extensions. The architecture of SMDA is presented in three orthogonal perspectives, including different abstract levels, modeling facets and involved technologies. SMDA is not only a high level methodology, but also able to convert requirement modeling to service implementation by semantic modeling techonologies. Guided by the SMDA method, the enterprise application development will be an iterative process which is business process focused and consists of service-oriented requirement analysis and modeling, services discovery and dynamic composition.SMDA method will be an important role in enterprise application development in the open and complex internet environment, which enhances the communication, gets rid of the heterogeneity, satisfy a variety of different users with dynamic and personalized requirements, and achieves the service-oriented efficient and agile development.Second, study the major techonologies in development process, including the requirement analysis and modeling, business process design and verification, services discovery and QoS-oriented global optimization approach for dynamic service composition.1) Propose a multi-ontology based requirement analysis and model method.Inspired by semantic ontology technology and domain engineering, we develop a multi-level ontology system, in which environment and goal ontology serve as the basis while organization, process and service ontology as business cores. Three types of different abstract level ontology:upper ontology, domain ontology and application ontology are developed to guide requirement analysis and modeling in reuse-oriented enterprise application development, and ensure the consistency of requirements. The goal oriented requirement modeling process starts with the organization structure analysis, focuses on the business process, and is finished when the service-based solution is created. By reusing domain model, the quality of requirement analysis and the efficiency of modeling process get improved.2) Propose an approach for business process modeling and formal verification.Combing with semantic ontology technology, based on UML (Unified Modeling Language) extensions, the approach supports visual modeling with multiple views according to different concerns in business process. The business modeling consists of three stages of refinements:1) whole abstract process; 2) declarative process; 3) imperative process. By adopting the concept of "environment ontology", the capabilities and behaviors of software can be specified by the effects of changing environment status. Based on the idea, the modeling related definitions and their formal semantics are presented. At last, an example of simplified products trading system demonstrates how to verify process model definition and model refinement specification with Alloy in details. The example shows that modeling business process approach and verifying the model specifications with Alloy can improve the model process flexibility and ensure models consistency.3) Propose an ontology-reasoning based goal-oriented services discovery approach.Services discovery is a key step in service oriented implementation. To deal with the problem of dynamic services discovery, we extend the SOA by leveraging the concept of abstract service type and service pools. By dynamically constructing the abstract service related service pool, a uniform service view is provided in the dynamic and heterogeneous environment. And based on this view, combined with ontology reasoning technology, a goal oriented service discovery method is proposed to convert the process model based service discovery problem into the problem of matching process goal with service goal. The ontology related with transaction process are used to discover satisfied candidates of service instances. And cases study show that this approach is able to construct constrained service pool using current ontology reasoning technology, effectively reduce the number of instances in the service pool, and facilitate the composition oriented and Qos global optimized service selection in the next step.4) Propose a QoS-oriented global optimization approach for dynamic services composition.Aiming at the problem of mass services composition, we present the services composition model and the QoS model. The services composition problem was cast to single goal multiple constrains problem. Based on the model, we propose an enhanced genetic algorithm named GOAQSC (Global Optimization Algorithm for QoS-aware Services Composition), which uses the strategies of high-fitness initial population and mutation with local searching to speed the convergence. The experimental results show that the proposed GOAQSC algorithm can return the near optimal solution more quickly.