Research On Technologies Of Web Services Dynamic Composition And UDDI Registration/Discovery

As a focus of IT industry and a very potential technology, Web Services defines a set of standards for inter-operating between applications on the web and can be described, published, discovered and invocated through the network. Athough Web. Services have already brough many conveniences for the integration inside and between the enterprises, the functionalities of some single web services are limited and it is not enough for application integrating among enterprise boundaries only with those simple-functionalities Web services. Therefore, we need to compose the existing single Web Services into some new Web Services so as to provide more sophisticated functionalities. As a Web Services registry center on the network, UDDI Registry takes on the task of service registration and discovery, plays an important role in the open, distributed and dynamic Web Services environment. But the current UDDI Registries are not so mature. The two discovery methods "discovery based on keywords" and "discovery based on dirll down" cannot meet some applications' needs. The main research contents include:A Web Services dynamic composition modeling method is proposed in this thesis. This modeling method is to represent a Web Service with one or more rules, and then use all kinds reasoning algorithms in AI to find the composition plan. A parameter-ontology is used to describe the semantic relationships among the inputs/outputs of Web service models in order to eliminate the possible semantic conflicts during composition process.This thesis introduces two Web Services dynamic composition algorithms. The main idea of the Non-backtrace BCA is to generate a composition plan by backward deduction according to the expected outputs given by end user. In order to avoid backtrace in the deduction process, an input closure may be computed with the inputs provided by end user. This algorithm is suitable for composing many times under the same inputs and the same models. The QoS is considered in the algorithm of Web Services optimal composition based on QoS. The key of this algorithm is using a deduced network to compute all the possible compositions and then take a predefined QoS model as a standard to select the optimal composition plan.Two automatic converstion algorithms are given for converting a composition plan into a composed Web Service. WSTC conversion algorithm is suitable forconverting the composition plan, which is generated from the Non-backtrace BCA, to a composed Web service. This algorithm is similar to the idea of Topological Sort algorithm. While WSDC conversion algorithm is a direct conversion algorithm which is suitable for converting the composition plan, which is generated from the Automatic compositon algorithm based QoS, to a composed Web Service.This thesis also presents a framework of Service registration and discovery in a domain-oriented UDDI Registry. Based on the current UDDI data model and specifications, this architecture uses an external database to support service registration and discovery based on service properties, service relationships and service constraints, reduce the returned services and so as to increase the satisfactory and accuracy of service discovery and meed all kinds of requirements of applications.The prototypes based on above approaches and technologies have already been implemented and applied in the machine-tools domains by a Shanghai Science Committee project "The Discovery and Appraisement of Domain-Oriented Manufactural resources".