【作者】 张霞；

【导师】 李德仁； 龚健雅；

【作者基本信息】 武汉大学 ， 摄影测量与遥感， 2004， 博士

【摘要】 Internet是信息发布与传播的一个重要途径,它一直处在快速发展之中。如今正在变为Web服务的媒介。Web服务系列技术标准的出现和发展正在引发信息技术的另一场浪潮。软件工业的发展经历了集中式——模块化——对象化——组件化的过程,目前正在向Web服务的方向发展。XML和Web服务技术的出现改变了Web应用程序的构造模式。XML具有强大的自描述能力,实现了数据与表现形式的分离,在Web信息共享及异构应用数据交换方面具有其它技术无法比拟的优点。Web服务包括一系列基于XML的标准技术集合。 Web服务的概念模型来自面向服务架构(SOA),Web服务是一种实现SOA的技术集合。面向服务架构(SOA)是一种软件构架概念模式,与以技术为中心的应用程序结构不同。面向服务架构由服务构成,最主要的特征是把服务的实现和服务的接口分离。服务具有一些特性,包括:服务是可以发现的并可以动态绑定的;服务是自包含的、模块化的;服务是可以互操作的:服务与服务消费者之间是松散耦合的;服务具有网络可寻址接口;服务具有粗粒度接口;服务的位置可以对服务消费者透明;服务是可以组合的;服务具有从错误状态恢复的能力等。SOA采用“发现、绑定和执行”模式,由服务消费者、服务提供者、服务注册中心、服务契约、服务代理和服务租约六个实体实现。同以技术为中心的应用程序相比较,服务从业务流程的角度看待技术,这和以技术驱动的业务处理相反。SOA解决了服务的描述与互操作等问题,使服务能紧密地同业务流程结合在一起,能够通过服务组合完成业务处理流程。 地理信息系统一直受到信息技术的影响,不可避免地正在和Web服务结合发展。开放地理信息联盟(OGC)提出的服务框架与互操作协议栈是地理信息系统向地理信息Web服务发展的开端,得到了许多地理信息系统(GIS)供应商的支持。越来越多的GIS供应商开始把其产品改装成Web服务,改变了以往各自采用私有架构方式的局面。随着地理信息Web服务的发展,将有越来越多的服务供应商把自己的地理信息Web服务发布在网络上。单个的地理信息Web服务提供特定的功能,如何组合网络上的地理信息Web服务完成单个地理信息Web服务无法完成的地理信息任务是值得研究的问题。本文针对地理信息服务组合的相关问题进行研究,具有一定的研究意义和价值。 从国内外发展现状来看,对地理信息服务组合的研究仍然处在初步阶段。需要研究的内容很广泛。本文所做研究工作是对地理信息服务组合理论的探索。主要研究内容包括地理信息服务组合的概念和模型、地理信息服务分类方式及其对地理信息服务组合的影响。 首先,从地理信息服务组合的概念出发,探讨适合地理信息服务组合的抽象模型。在分析总结SOA、ISO和OGC的服务组合概念基础上,将服务组合的更多还原

【Abstract】 Internet is an important intermedium of information issue and transmission. It has been developing with a high speed. Nowadays, it has been becoming a media of Web Services. Web-oriented services develop quickly and services fields become more widely. The adventure and development of Web Services technologies and specification bring another information technology storm. The development of software industry has experienced integration, modularization, object orientation, and component. Now it is developing towards Web Services.With the adventure of XML and Web Services, Web application developing mode has been changing. XML is greatly capable of self-description. It separates data content and data presentation. It has been proved that XML is very suitable of implementing Web information share and heterogeneous application data exchange. Web Services refers to a set of XML specification and technologies which partly implement service-oriented architecture (SOA).SOA is a special kind of software architecture that has several unique characteristics. Different from technologies-oriented architecture, SOA deals with services. The most important aspect of SOA is that it separates the service’s implementation from its interface. The unique characteristics of SOA includes as the following. Services are discoverable and dynamically bound; Services are self-contained and modular; Services stress interoperability; Services are loosely coupled; Services have a network-addressable interface; Services have coarse-grained interfaces; Services are location-transparent; Services are composable; and Service-oriented architecture supports self-healing. SOA support the "find, bind, and execute" paradigm. SOA consists of six entities of service consumer, service provider, service registry, service contract, service proxy and service lease configured together to support the find, bind, and execute paradigm.Compared with technologies-oriented application, services treat technologies from the point of business processes, which is opposite from technologies-driven business processes. SOA can resolve the problem about services description and interoperation, which make it possible to combine services with business processes tightly together. Therefore, business processes can be accomplished with services combination.Geographic Information System (GIS) is always influenced by the development of information technologies. Nowadays, it is unavoidable to develop towards Web Geographic Information Services (GIServices). OGC has proposed OpenGIS WebServices Architecture and interoperate protocol stack which have been supported by many GIS providers. More and more provider has been packaging their products into Web Services, which change the situation of developing WebGIS with private technologies. More and more will be put on the Web. Single Web GIService provides particular function. How to combine Web GIServices to accomplish Geographic Information tasks that can not be accomplished by on service is worth studying. This paper is focused on GIServices combination, which is significative.According to the development status of inland and abroad, study on GIServices combination is still at primary stage. Study of GIServices Combination involves many fields some of which is explored in this paper. This paper focuses on GIServices Combination concepts and models, GIServices taxonomy system and its influence on GIServices Combination.Firstly, the concepts and abstract model of GIServices Combination are studied in this paper. According to the service combination concepts of SOA, ISO and OGC, services combination are classified as discrete combination and chain combination. It is proposed that service combination can be recursive, which means that composite services can also be combined. It is called hybrid combination. In consideration of service final consumers - users, it is believed that chain combination is the main services combination form. This paper focuses on the study of chain service combination based on workflow theory. Service combination (chain form) is related with workflow. Concepts of service item and service instance are brought forward according to work item and activity concepts in workflow. Service combination three-dimension perspective is produced according to that of workflow. On the other hand, there exist differences between service and task, service instance and activity, service combination and workflow. It is proposed that service combination should be built based on users tasks, which is called as tasks/activities-oriented service combination.Petri Net is very suitable for workflow model. Rachid Hamadi and Boualem Benatallah have proposed to model web service control flow and web services combination with Petri Net. The model they built is extended and a high-level service combination model based on Petri Net is produced in this paper. Service combination is directly mapped into a Petri Net. Services are modeled by transitions and the state of service combination is modeled by places. The arrows between places and transitions are used to specify causal relation. Internal activities of service are hided and a service is modeled as a single node in the net, which makes the combination relationship clear and the control of combination behavior easy. Based on the high-level service combination model, five chain forms, sequence, unordered sequence, alternative, parallel without communication and iteration, are depictedwith Petri Nets.Secondly, GIServices taxonomy and its influence on GIServices Combination are studied with emphatically. To implement tasks/activities-oriented geographic information services combination, correspondence between users tasks and service types and matching between activities and service instance which is concerned with geographic information process semantics should be established. Semantics of geographic information service is concerned with services taxonomy.Two existing GIServices taxonomy are discussed. One is proposed by ISO/OGC based on Extended Open Systems Environment. The other is based on tasks taxonomy. The disadvantage of the former is that it is not suitable to built services combination by users. The disadvantage of the latter is that service semantic granularity is simplex. To overcome these shortcomings, a multi-level services taxonomy system based on users and tasks is proposed. Geographic information tasks are classified at different users level including common market users, high-level expert users and low-level expert users. Consequently, the GIServices taxonomy consists of high-level taxonomy and low-level taxonomy.GIServices taxonomy has influences on service metadata that is the basis of corresponding tasks with services type and accomplishing service instance matching. Service metadata is studied in this paper. Service metadata of service in users and tasks oriented GIServices taxonomy is created based on the services taxonomy and Service Information Model (SIM) of OGC Web Services (OWS).Different GIServices combinations based on different services taxonomy are analyzed and compared with use cases. It can be concluded that users and tasks oriented GIServices taxonomy system is more suitable for tasks/activities-oriented service combination. It can satisfy the task process requirements of different users.On the basis of GIServices model and taxonomy, Spatial Analysis Service is investigated in this paper as the second theme. Many processes can be done on geographic information. Among them spatial Analysis is one kind of geographic information process and always used by users. In users and tasks oriented GIServices taxonomy system, spatial analysis service is one basic service of high-level and service corresponding with specific spatial analysis process is the basic services of low-level. Since existing services taxonomy is not based on users level and tasks classification, study on spatial analysis service is insufficient, which makes it valuable to study spatial analysis service in this paper. The study involves Spatial Analysis Service description (service metadata), service interface and its access mode, sevice access protocol and service implementation.In this paper, spatial analysis service is studied and implemented based on XML and Web Services. Firstly, Spatial analysis service metadata and service interface aredefined with XML. Access schema to service interface is given. To define request and response protocol which is used to access spatial analysis service, two geographic feature data schemas are defined firstly. One is called direct coordinate feature, the other is called indirect coordinate feature. According to feature data types of request and response, the interaction between service consumer and service is sorted as direct-direct, direct-indirect, indirect-direct and indirect-indirect interaction. They are all modeled with Petri Nets. The request/response message schemas are then defined with XML.Then, a physical architecture of implementing spatial analysis service based on Java Web Services, J2EE is proposed. Then spatial analysis service implemented is combined with feature provide service and feature render service for a specific use case.Finally, the performance of spatial analysis service implemented with XML and Web services is evaluated through comparing it with another service which is accessed with bytes stream protocol. In the comparing experimentation, same features are selected to conduct same spatial analysis operation with two different services. The result shows that the volume of data in XML schema is obviously greater than that in bytes stream format and the time consumed to access spatial analysis web service is obviously longer than that of comparing service. Therefore, the performance of service implemented with XML and Web Services should be improved. It is suggested that the data volume can be reduced greatly with indirect coordinate schema, which can improve the service performance.Summarily, two themes are involved in this paper. Some primary research work about GIServices combination is conducted, and spatial analysis service is studied and implemented in this paper. Study on the latter is based on that on the former. Main contributions and innovations include the following. CD Problems of GIServices combination are studied. A Petri Net-based high-level services combination model is proposed. (2) Users and Tasks Oriented GIServices taxonomy is brought forward. Service metadata is built based on this taxonomy. (3) Spatial analysis service interface, interface access schema and sevice access request/response message schema are defined. (4) Spatial analysis service physical architecture experiment is conducted based on Web services and J2EE. A spatial analysis service instance and its combination with other GIService instance are implemented.All the work in this paper represents an attempt to the research and application of GIServices combination, which can be useful for later research.更多还原