Research On Web Service Matching Based On Semantics And Dynamic QOS Model

Because of various superiority and broad application prospect,web service is praised and canonized by many organizations and companies since the day of its appearance.Many companies introduce web service into their commercial procedures and benefit from it.But with the development of web service,the amout of services available on the internet is becoming larger.So how to find the service that can satisfy the consumer's function request from a huge amount of services-web service matching turns a problem which must be solve immediately.At present, web service matching mechanism of UDDI is based on key words,which leads to low recall and precision ratio.With large numbers of web services emerging, there will be more and more services with similar function .But these services may be greatly different in their nonfunctional attributes,such as charge, reliability and stability. The traditional match mechanism lacks Web service quality supports so that it can't realize the discovery of best service. Consumers may have different demands about the nonfunctional attibtutes.Therefor,how to find the services that the requester regard best according to his demand is also a problem which we should take into account in the course of service matching.This paper introduce ontology technology into the web service and give a web service matching model based on semantics against the drawback of traditional matching mechanism.Then we propose a multi-layer and finely granular matching strategy and improve the arithmetic of semantic similarity computation. The specific work as follows:First,we do some research on how to definite and describe the web service.The semantic description model of web service is created in order to support the quality of service.We realize that the semantic description can be used in UDDI by means of registering the semantic and QOS description into UDDI in the form of TModels.Second,Being opposite to the disadvantage that traditional model can't meet the demand of semantic matching,we bring forward a model based on semantics and dynamic QOS(quality of service) and design its hiberarchy,in which the model is devided into four layers and several modules such as service publishing,service discovery, OWL-S/UDDI mapping and so on. Third,on account of how to find the best service from the services with same function according to the QOS,this paper give a new QOS description model.We bring forward an effective QOS model and design the evaluating arithmetic of QOS measurements considering synthetically the history statistical information of QOS.This lay the foundation of the last research on service description and matching.Fourth,some research on matching between the advertizing and the requesting services is done according to the intelligent and efficient requirement.We propose a multi-layer and finely granular matching strategy,which is to return consumer's satisfied service by means of matching category,basic description information,I/O and QOS of services,Besides,we improve the arithmetic of semantic similarity computation.Last,we simulate the function of the model and carry on the verification of validity and function test to our algorithms. Results indicate that the model this paper put forward can support semantic matching of web service well and the arithmetic can improve the recall and precise in some degree.