| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 95 === Service-oriented architecture (SOA) is a concept of program architecture, can offer flexible, and efficient, and IT environment that provides integrated information. The web service is basing on the concept of SOA, is a standardized, distributed computing component which has been adopted in industry. The utilization of web services can achieve the software with the advantage of highly reusable, loosely coupling software architecture, distributed computing and dynamic service providing on different platforms. Service-matchmaking component is one of the important keys for SOA system development. Current service-matchmaking methods mainly use the method of key words and service category search in an UDDI (Universal Description, Discovery and Integration) for the suitable WSDL. However , since the information of WSDL is so limited, it is unable to get useful information from WSDL. Therefore, one solution is to introduce the ontology of semantic web to enhance web services description.
Many researches emphasized the accuracy of service-searching based on the reasoning of description logic’s, but most issues are subject to calculating under static searching mode which doesn't involve the ability of self-adoption. The context of this thesis combines self-adapted decision-making mechanism of voting with description logic’s (DL) subsumption reasoning. It could produce customized and precise search algorithm. The main idea of this self-adapted voting decision-making mechanism is based on the application of evolution computation as the learning algorithm; then to do the simulation by means of neural network to perform the decision behavior of the end-user. Afterwards the compliance is checked by a voting mechanism to increase its accuracy. The searching system will automatically simulate the end-users within the criteria of how to perform the decision-making which involves the ability of self-auditing. Therefore it could be responded and send feedback to the end-user with enough and accurate information.
|
|---|
