【作者】 钱海忠；

【导师】 沈苏彬；

【作者基本信息】 南京邮电大学 ， 信息网络， 2013， 博士

【摘要】 随着面向服务计算模式深入应用,语义万维网服务作为万维网服务与语义万维网两大技术融合的一个领域成为研究热点。语义万维网服务是目前实现“云计算”按需服务的关键技术。传感器万维网、物联网以及“云计算”等诸多新兴领域的兴起,使得语义万维网服务若干关键技术的研究不仅具有重要的理论意义,也具有实际的应用价值。近年来国内外针对语义万维网服务中的万维网服务语义标注、万维网服务注册与发现以及万维网服务自动组合等若干关键问题取得了许多有价值的研究成果,但是这些问题还有待深入研究。论文针对这几个关键问题展开了深入的研究工作,并把一些研究成果初步应用到传感器万维网中。论文主要贡献可归纳为以下几个方面：(1)提出了一种基于语义词典的万维网服务语义标注方法。现有语义标注方法的前提条件是各方需要使用共同认知的领域本体库,但是在万维网这类互联网环境下,这个前提条件是难以做到的。本文提出了一种基于语义词典的服务语义标注方法,解决服务语义标注方法对共同认知领域本体库依赖的问题。仿真实验表明基于语义词典的服务语义标注方法,能够提高服务发现的准确率,同时该方法容易被用户使用,并且与现有的一些万维网服务语义标注方法(例如SAWSDL)相兼容。(2)提出了语义万维网服务检索树的构造算法,解决了注册万维网服务信息的有效组织问题,同时给出了基于语义万维网服务检索树的服务语义快速发现算法。语义万维网服务检索树可部署在一个服务注册中心,把那些在语义层面上具有相同输入参数集合的服务聚集在一个树干节点之上,而对应的每个服务作为该节点的叶子节点,以此缩小万维网服务发现算法的搜索空间。本文采用元素检索矩阵的方法,快速查找到语义万维网服务检索树的树干节点：首先在服务注册过程中给树干节点赋予一个唯一值并建立该值和节点之间的索引；然后利用树干节点中存储的所有参数集中的元素(万维网服务的输入参数或是它们的交集)建立一个元素检索矩阵,树干节点的值可以通过查询元素检索矩阵得到并快速找到相应节点；最后服务发现算法将在此节点依据请求服务的功能属性查找在语义上匹配的服务。虽然语义万维网服务检索树的构造要花费一定的时间,但是该过程可作为一种预处理的手段,减少了服务发现阶段的语义计算时间。实验表明基于语义万维网服务检索树的服务发现算法与同类服务发现算法中有较快的响应时间。(3)提出了一种基于状态演算理论的服务动态组合优化算法。本文把万维网服务动态组合过程看成是一个状态演算过程。用户请求的输入参数集合作为初始态,用户想要获得的结果集作为终止态,而每个万维网服务都看成一个能促使状态变迁的动作。注册中心的万维网服务不需要按照传统算法进行形式化转换,服务组合算法根据当前状态的信息(用户提交的和中间过程中产生的),采用服务快速发现算法去匹配和发现满足条件的原子服务,直到状态变迁到满足结果集为止。整个状态的变迁过程中可以记录各个原子服务之间的依赖关系和执行先后顺序,因此可以简化服务组合方案。与同类算法进行仿真实验比较,结果表明本文提出的算法具有较好的时间效率。(4)把基于语义万维网服务检索树的服务注册与发现算法应用到传感器万维网领域,解决了传感器万维网中服务的注册和基于语义的服务发现问题。SWE采用OGC目录服务来管理和发现传感器万维网服务,但是这种服务发现还是处于语法层面。本文结合观测与测量O&M (Observation and Measurement)标准,对O&M中每个评测项用具体参数集合进行细化。例如,用(一氧化碳、二氧化硫)来细化观测空气污染程度,以满足语义万维网服务检索树的构造性质,便于利用语义万维网服务检索树来管理和发现传感器万维网服务。本文从实际应用角度研究了当前语义万维网服务中若干需要解决的问题,并提出了相应的解决方案。这些研究成果在推动语义万维网服务在实际中的深入应用具有一定的理论意义和应用价值。更多还原

【Abstract】 Along with the further application of Service Oriented Computing, The technology of Semantic web Services as the integration of the web services and semantic web becomes a hot topic. The Semantic web Services is also the key technology for on-demand service of cloud computing. With the emergences of some new fields, such as Sensor web, Internet of Things, and cloud computing, the study of key technologies of Semantic Web Services not only has theoretical significance, but also has practical application values.Although a number of valuable research results on the key techniques of semantic annotation of web service, service registry and publishing as well as the service composition process generation have been achieved, those need to be further studied. The main contributions of this thesis are summarized as follows:(1) A new semantic annotated method based on the semantic dictionary is proposed. The prerequisite of the existing semantic annotation method for web services is that all parties need to use the common-known domain ontology. But this prerequisite is difficult to achieve in the Internet environment. A method is proposed in this thesis based on the semantic dictionary, such as Hownet, WordNet, to solve the issues of semantic annotation algorithm dependent on common-known Ontology. The experiment results have shown that the method can improve the accuracy of service discovery. The method also can be used easily by users and is compatible with some existing semantic annotation method (such as SAWSDL).(2) An algorithm of constructing SWSQT (Semantic Web Service Query Tree) is proposed to organize the information of registered web services. A quick web service query algorithm based SWSQT is also proposed. SWSQT can be deployed in a service registry center. All the web services that have the same semantic input parameters will be organized into one trunk node of SWSQT, and the each related web service will be attached to the node as a leaf node. So the searching space of algorithm can be reduced. The trunk node can be located by means of building and calculating the corresponding query matrix. Firstly, each trunk node is assigned a unique incremental value and an index is also established by means of the value during service publishing process; then all of the elements of the trunk node are organized in a query matrix and the value of each trunk node can be got by querying the matrix; lastly, the algorithm of service discovery only discovers the possible matched services in the trunk node. Although the process of constructing SWSQT will consume some time, it can be act as a pre-process and the large number of ontology reasoning time can be reduced in the following service query process. The experiment results show that the proposed algorithm based SWSQT has faster response time than that of similar algorithms.(3) A web service composition optimization algorithm based on the state calculus is proposed. The web Service dynamic composition process can be regarded as a state calculus process. The input parameters of users as the initial state, the results that the user wants to achieve as a final state. Each web Service is as an action, which can change web service compositing state. The registered web services do not need to be formalized conventionally. The composition process discovers、matches the atomic web service by the information, which is provided by user and generated in process, and changes the compositing state, until the state changes to meet the final state. The dependencies and order of execution of all atomic web services will be recorded in the composting process, so the choreography of the services can be achieved easily. By comparing with the similar algorithms, the experiment results show the proposed method has better time efficiency.(4) The proposed algorithms of web services registration and discovery based on SWSQT are applied directly to the sensor web, the problems of the web service registration and discovery based semantic are intended to be resolved. For the SWE uses the OGC Category services to manage and discover the services, so the matching of services is still based on syntax. The items of O&M (Observation and Measurement) are redefined by a collection of annotated semantic parameters to satisfy the requirements of constructing SWSQT, for example, it can use the collection of carbon monoxide, sulfur dioxide to represent the air pollution observations, and the SWSQT also storage each URN (Uniform Resource Name) of sensor instance or services.In conclusion, the current issues of the semantic web services have been researched in the view of practical applications in this thesis. Some solutions are proposed to solve the problems. These works may have certain theoretical significance and practical application values.更多还原