Research On Web Services Discovery And Adaptation Based On Interface And Behavioral Semantics

Today we are experiencing a major paradigm transformation in the way that software applications are designed, delivered and consumed. Service Oriented Computing (SOC) is the newly-emerging and highly-promising paradigm for distributed computing and software engineering. A series of techniques are incorporated, guiding the way to construct distributed, loose-coupled, on demand enterprise application systems where services are reused as basic elements.The core principle of Service Oriented Computing is to reuse services, and the key is to find out the services conforming to the requirement of application systems. On the other hand, with the wide adoption of Web services supported by continuously improved technologies, standards and platforms, the published services have mushroomed over the Internet, which brings us an urgent challenge to locate target services in a quick, accurate and efficient way.The research on Web service discovery comprises of a series of crucial issues, and in this thesis, we focus on the following three key issues:Web services description model and meta-data management, Web services discovery based on interface and behavioral semantics, and semantic adaptation of candidate services. Major work and contributions include:First, we propose a formalized model of Web services and approaches to manage Web service instances. The model of Web services consists of a semantic interface model for atomic services and a semantic behavioral model for composite services, where the dependency relationships among messages are captured. It is further formalized withπ-calculus, which will be used in the analysis of service behaviors. We then propose to manage a large number of Web service instances by adaptive Back-Propagation neural network based classification and inverted indexing, which can facilitate service discovery based on interface semantics.Second, we propose approaches of service discovery oriented to atomic services based on interface semantics and that oriented to composite services based on behavioral semantics respectively. For service discovery based on interface semantics, the dependency relationships among messages are considered in the matchmaking of single service to reduce false negatives, and instant compositions of multiple services are explored to increase candidate matches. And the most important, the efficiency of the two-phase service discovery based on interface semantics are improved by the established inverted indexing. For service discovery based on behavioral semantics, we propose 6 notions of services behavioral equivalence with implication relations and a derivative approach to analyzing the behavioral equivalence for Web services automatically according to the given notions. The approaches as well as the application scenarios are illustrated through examples and experiments.Third, we propose a methodology for the development of adaptors capable of making two incompatible services interoperate not only successfully but also correctly at semantic level. We propose to generate adaptor specification according to the descriptions of two services to be adapted step by step. Message mapping is first generated to specify the mapping relations among input/output messages of the two services. Action mapping is then generated to specify the mapping relations among actions of the two services, and meanwhile, treatment for non-mapping is generated to specify how to deal with the messages without a mapping by considering their semantics in the interaction of two services. Starting from the adaptor specification and two service descriptions, an incremental derivation of a concrete adaptor is given.Fourth, we propose a framework of service discovery system and implement a prototype. We outline the main functions, components and core processes in a service discovery system. According to the proposed framework, we design and implement a service discovery prototype named Jtang service community, in which we realize the classification of Web service instances, service discovery based on matchmaking of interface semantics between atomic services, and behavioral semantics equivalence for composite services.