基于社会网络的语义Web服务发现与组合研究

发布时间：2018-04-14 21:18

本文选题：语义Web服务 + 社会网络　；参考：《南开大学》2014年博士论文

【摘要】：基于服务架构(SOA)的语义Web服务,通过服务重用,具有自包含、自描述、跨平台等优点,在电子商务、企业应用集成等领域具有广泛的应用,已经成为工业界和学术界关注的焦点。语义Web服务的研究重点是服务的发现和组合。服务的发现主要涉及服务的查准率和查全率,服务的组合主要关注服务的自动组合。由于社会网络能够充分利用网络实体问的上下文关系,发现网络实体,因此研究者把语义Web服务和社会网络相结合提出语义Web服务的社会网络,用来提高语义Web服务的发现效率。但是社会网络的引入带来了新的问题：如何组建语义Web服务的社会网络；如何在组建的语义Web服务网络中进行语义Web服务的查找；如何优化所组建的语义Web服务网络；如何在语义Web服务网络中进行服务组合等。本文围绕这些问题,对社会网络在语义Web服务中的应用展开研究,创新点如下： (1)提出了一种有向加权的语义Web服务社会网络。现有的语义Web服务社会网络大多为无向无权的服务网络,这类方法只能从定性的角度研究语义Web服务社会网络。针对该问题,本文将语义Web服务及其之间的关系形式化为一种有向加权的服务网络。该方法以语义Web服务的集合为结点,以服务集合之间的连接为有向边,以服务集合之间连接的强弱关系为权重,建立语义Web服务的社会网络。同时研究了服务网络中的领域本体库、服务请求者、服务提供者和服务注册中心的交互机制,用于对服务网络进行更新。以商品服务流程为例进行实验,结果表明该方法所建服务网络能较好的描述语义Web服务的各项特征。 (2)提出了一种基于紧密度的层次聚类语义Web服务社会网络组建方法。现有的语义Web服务聚类方法难以确定聚类结果中簇的数量,在进行聚类时需要人工设置较多的参数,导致这些参数只适用于特定的数据集。针对该问题,本文提出了一种基于紧密度和互连度的层次聚类方法。该方法利用语义Web服务之间的相似度进行聚类,形成语义Web服务社会网络的结点,通过控制结点的紧密度和互连度调整结点中语义Web服务的数量,细化语义Web服务的分类。同时为了提高服务网络中语义Web服务的查找效率,本文研究了两阶段的倒排索引方法,通过本体-服务对和服务-结点对的两阶段倒排索引进行语义Web服务的发现。利用OWL-TC4数据集进行实验,结果表明改进的层次聚类方法提高了语义Web服务的聚类准确率,两阶段倒排索引方法减少了语义Web服务的查找次数,提高了查找效率。 (3)提出了一种基于补偿代价的动态服务网络结构优化方法。现有的研究方法大多只考虑了在一个时刻时语义Web服务组成的社会网络,没有考虑不同时刻服务网络结构的变化,因此不适应语义Web服务的动态性。针对该问题,本文提出了一种基于时间序列的网络结点动态优化方法。该方法将服务网络的时序动态性和时刻静态性用时间序列的方式表示,服务网络在时间序列的每一个时刻是静态的,同时又随时间序列动态变化。语义Web服务的补偿代价是指语义Web服务在更新、失效等情况下,服务网络为了实现该语义Web服务所具有的功能而需要付出的代价。通过比较语义Web服务在服务网络结点更新前和更新后的补偿代价,从而判断语义Web服务网络结点是否需要更新,动态确定语义Web服务的所属结点。利用不同时间段的QOS数据集进行实验,结果表明该方法可以有效减少网络结构的变动次数,得到了稳定的服务结点,提高了服务的查找效率。 (4)提出了一种基于服务网络结构的语义Web服务组合流程验证方法。现有的语义Web服务组合流程大多采用形式化方法进行验证,随着语义Web服务数量的增多,服务组合的验证效率越来越低。针对该问题,本文提出一种基于服务结点可达概率的语义Web服务组合验证方法。该方法以服务组合流程中的语义Web服务所在结点为研究对象,将语义Web服务的组合流程转化为服务结点的可达路径,通过有向边的权重确定两个服务结点的可达概率,结合服务网络的中心度和关键路径,判断是否需要对服务组合进行验证。在服务网络的基础上利用仿真数据进行实验,结果表明该方法可以有效减少语义Web服务组合的验证次数,提高了服务组合的验证效率。综上所述,本文采用社会网络的方法研究了语义Web服务的发现和组合,并取得了一定的研究成果,对促进语义Web服务社会网络的研究具有重要的理论意义和应用价值。
[Abstract]:Based on the service architecture (SOA) of the semantic Web service, through service reuse, with self-contained, self describing, cross platform advantages, in e-commerce, has been widely applied to enterprise application integration and other fields, has become the focus of attention of academia and industry. The research focus is the semantic Web service discovery and composition of services the service discovery service. Mainly related to precision and recall, automatic combination mainly focus on service.
Because the context of social network can make full use of the network entities, discover network entities, so researchers have semantic Web services and social networks is proposed based on semantic Web service and social network, to improve the semantic Web service discovery efficiency. But the introduction of social networks has brought new problems: how to build a semantic Web service society how to carry out the network; semantic Web services in semantic Web service network in the search; how to optimize the formation of semantic Web service network; how to carry out the service composition of semantic Web service in the network. This paper will discuss these problems, research on the application of social network in the semantic Web service and innovation are as follows:
(1) proposed a weighted semantic Web service social network. The existing semantic Web service social network are undirected and unweighted network of services, only this kind of research methods from the perspective of qualitative semantic Web service social network. Aiming at this problem, this paper will form a relationship between the service and the semantic Web as a weighted network. This method based on the set of semantic Web services as nodes, to connect the service set between the directed edges, the set of services connected between the strength of the relationship between the weight, establishment of semantic Web service and social network. At the same time on the domain ontology library services in the network, service request the interaction mechanism, service provider and service registration center, service network for goods and services to be updated. Process experiments, the results show that the method of service and the neural network can describe the semantics of Web. All the features of the service.
(2) proposed a method of building the close degree of hierarchical clustering of semantic Web service based social network. Semantic Web service clustering method is difficult to determine the number of clustering results of existing cluster in the clustering parameters in hand more, leading to these parameters are only applicable to a particular data set. To solve this problem, this paper this paper proposes a hierarchical clustering method based on degree of closeness and interconnection. The method uses similarity between semantic Web services clustering, a semantic Web services social network node, by the number of compactness and the interconnection degree adjustment of semantic Web service nodes in the control node, classification of semantic Web services. At the same time in order to find the efficiency of semantic Web services to improve the services in the network, in this paper the inverted index method in the two stage, the service of ontology and services - nodes of the two stage inverted index The discovery of semantic Web services. Experiments on OWL-TC4 dataset show that the improved hierarchical clustering method improves the clustering accuracy of semantic Web services. The two stage inverted index method reduces the number of semantic Web services lookup and improves the lookup efficiency.
(3) propose a method based on the cost of compensation of the dynamic service network optimization method research. Most existing methods only consider the semantic Web service composition at a time when the social network, not considering the variation of different time service network structure, so it is not adapt to the dynamic of semantic Web services. Aiming at this problem, this paper proposes a network node dynamic optimization method based on time sequence. The dynamic and static timing of the service network represented by time series method, service network at every moment of time series is static, but also dynamic changes with time sequence. The cost of compensation is the semantic Web service the semantic Web service in the update, such as the failure cases, in order to realize the service network has the function of semantic Web service needs to pay the price. Through the comparison of semantic Web service in the service network The cost of compensation nodes before and after updating, in order to determine whether the semantic Web service network node needs to be updated dynamically determine the node belongs to semantic Web services. In the experiments with QOS data in different time periods, the results show that the method can effectively reduce the network structure change times, obtain a stable service node, improve the the efficiency of searching service.
(4) propose a semantic Web service composition process verification method of service network structure based on semantic Web service composition process. Most of the existing verified using formal methods, with the increase in the number of semantic Web service, service composition verification efficiency is more and more low. Aiming at this problem, this paper proposes a semantic Web service composition verification method of service node based on reachable probability. The method on the service composition process in the semantic Web service node as the research object, the feasible path of semantic Web service composition process into the service node, determine the reachable probability two service node to the edge weights by combining, centrality and critical path network service the judge whether need to verify the service composition. By using the simulation data based on service network on the experiment results show that this method can effectively reduce the semantic We The number of validation of the B service composition improves the efficiency of the validation of the service portfolio.
To sum up, this paper studies the discovery and combination of semantic Web services using social network approach, and has achieved some research results. It has important theoretical significance and application value to promote the research of semantic Web services social network.

【学位授予单位】：南开大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】