节点文献

语义Web服务发现关键技术研究

Key Technology of Semantic Web Service Discovery

【作者】 张杨

【导师】 刘发贵;

【作者基本信息】 华南理工大学 , 计算机应用技术, 2014, 博士

【摘要】 面向服务的计算SOC是一种分布式的计算模式,它以服务作为构成分布式系统的基本要素。Web服务发现是Web服务研究领域的一个非常重要的组成部分,它是实现服务共享和重用的前提条件。然而由于Web服务缺乏机器可理解的语义信息,无法实现自动化服务发现。为了克服Web服务的不足,随着语义Web的发展,语义Web服务应运而生。语义Web服务是语义Web思想在SOC上的应用。语义Web服务发现克服了Web服务发现的缺点,实现了对服务的自动化发现。当前语义Web服务发现在学术界是一个研究热点问题。高效准确的语义Web服务发现对于语义Web服务的后续环节如服务组合、服务调用等具有重要意义。本文针对语义Web服务发现中的关键技术问题,主要的研究内容和创新点如下:1.融合基于逻辑推理方法和基于非逻辑推理方法,提出了一种面向服务功能的语义Web服务匹配方法。本文所研究的面向服务功能的语义Web服务匹配的主要思想是使用混合方法对服务请求者/服务提供者的IO参数进行本体概念匹配。主要工作如下:第一,定义了新的本体概念之间的关联关系分类,将概念之间的关联关系的分成五种类别;第二,根据概念关联关系的不同类别结构特点,提出了不同的相关度计算函数;第三,借鉴二分图最优匹配算法的思想,提出了基于二分图匹配的语义Web服务算法。2.利用图匹配的思想,提出了一种面向服务过程的语义Web服务匹配方法。本文提出了一种基于图的方法来解决面向服务过程的服务匹配问题,主要工作如下:第一,针对服务模型中的原子过程,利用图核函数的方法,设计了计算原子过程相似度的方法。第二,针对服务模型中的复合过程,设计了将OWL-S服务过程模型转换成OWL-S服务模型图表示方法,同时提出了将OWL-S服务模型图分解成为行为路径图的方法,定义了行为路径接口。第三,利用图核函数的思想,将计算两个行为路径图的结构相似度问题转化成为计算行为路径子图相似度的问题,提出了服务过程模型结构相似度的计算方法,并据此提出面向服务模型的语义Web服务匹配算法。3.针对各种类型的QoS参数,设计了相似度计算方法,提出了一种基于QoS的语义Web服务选择方法。本文提出了一种基于QoS的相似度进行单个服务选择的方法,主要的工作如下:第一,定义了服务请求者的QoS请求模型和服务提供者的服务发布模型,这两个模型能够很好地描述服务的QoS,并且考虑了用户的期望,具有一定的灵活性;第二,提出了一种基于QoS的服务选择方法,提出服务选择分为服务过滤和服务QoS相似度计算两个步骤;第三,提出了基于维度匹配和值匹配进行服务过滤的方法;第四,提出了一种新的计算QoS参数相似度的计算方法,它能够处理任何类型的数据。4.融合SON和超级节点思想,提出了一种基于三层结构的语义Web服务发现架构,着重研究了SON的管理问题。针对基于非结构化P2P网络的语义Web服务发现系统存在的问题,本文提出了一种适用大规模分布式系统的语义Web服务发现系统架构,主要工作如下:第一,提出了一种基于三层结构的语义Web服务发现架构,该架构的底层结构是非结构化P2P网络,第二层的语义覆盖网络(SON)建立在服务类别相似的节点基础之上,不同SON的超级节点之间存在逻辑连接构成第三层网络结构。第二,提出了基于SON的语义Web服务发现系统模型,定义了语义Web服务发现系统的功能。第三,研究了服务节点加入SON的组织方法和SON的构建和管理,提出了节点加入SON的策略和SON的构建和管理算法。

【Abstract】 Service Oriented Computing is a paradigm of distributed computing, which buildingblock is service. Web service discovery is very important in SOC research area. It is one ofthe prerequisites of implementing service sharing and service reusing. Lacking of semanticinformation, web service cannot achieve the goal for discovering services automatically. Toovercome the deficiency of web service above, semantic web service emerges with thedevelopment of semantic web. Semantic web service is the application of ideas of semanticweb. Semantic web service discovery realizes the automatic discovery of services.Currently, many researchers give their attentions on semantic web service discoveryproblem. Semantic web service discovery with high performance and precision is vital to thesubsequent sessions of semantic web service, such as service composition and serviceinvocation. This thesis aims to study key technology of semantic web service discovery. Themain contents and contributions are listed as follows:1. A semantic web service matchmaking method oriented on service functionality isproposed with the fusion of logic-based and non-logic based method.The main idea of this matchmaking method is to apply combination of logic-based andnon-logic based method to match the IO parameters of service requestor/service provider.Firstly, it defines a novel classification of semantic relevance relations between ontologyconcepts. We distinguish five types of semantic relevance relations. Secondly, it designsdifferent semantic relevance computation formulas based on the characteristics of differentsemantic relevance relations above. Thirdly, it designs a semantic web service matchmakingalgorithm based on bipartite graph borrowing the idea of bipartite optimal matchmaking.2. A semantic web service matchmaking method oriented on service process model isproposed using graph matchmaking.This dissertation develops a service matchmaking method based on graph matching forconsidering OWL-S service model. Firstly, it designs a method for computing atomic processsimilarity using the idea of graph kernels. Secondly, it designs a method for decomposing theOWL-S Service Model into OWL-S Service Model graph, and a method for decomposing OWL-S Service Model graph into behavior path graph. Additionally, it defines a behaviorpath interface for behavior graph. Thirdly, it devises a method for computing structuralsimilarity of compound process by transforming the similarity of two behavior path graph tosimilarities of corresponding behavior path sub-graphs. At last, this dissertation draws thesemantic web service matchmaking method oriented on service process model.3. A QoS-based semantic web service selection method is proposed based oncalculation of various types of QoS parameters.This dissertation designs a single service selection method based on QoS similarity.Firstly, it defines a QoS request model for service requestor and a QoS publication model forservice provider. These two models, with consideration of user’s expectation, are flexibleenough to describe the QoS information of service requestors and providers. Secondly, itproposes a two-step service selection method, which consists of service filtering and serviceQoS similarity computation. Thirdly, it designs the service filtering method. Fourthly, wedesign a novel computation method of QoS similarity.4. A SON-based semantic web service discovery system framework is proposed withthe fusion of SON and super-peer.This dissertation designs a semantic web service discovery system framework foradaption to large scale distributed systems, eliminating the shortcomings of unstructured P2Psystems. Firstly, it proposes a three-layer semantic web service discovery system framework,in which the1stlayer is a unstructured P2P network, the2ndlayer is semantic overlay network,and the3rdlayer is constructed with super-peers in each SON. Secondly, it proposes asemantic web service discovery system model and define the functionality of it. Thirdly, itstudies the organization and management of SON.

节点文献中: