Research On Technology Of Semantic Web Services Composition Oriented Information Sharing In Chinese Railway

Information can not be shared in Chinese railway enterprise because information systems are platform-different. Web service can integrate platform-different systems, and the technology of service composition can widen and deepen the use of information. Now this technology is hotpot in the field of web service and most crucial problems have not be solved yet, so research on this technology which is oriented information sharing in Chinese railway is important. Crucial problems of this technology include the classification of semantic web service, the model of semantic web service composition, the methods of composing semantic web service and the method of running composite semantic web service.The capability of semantic web service is embodied by Input, Output, Precondition, and Effect. The last two are the most important in the four parameters. Most of the semantic web services of railway have only I/O parameters, so output parameter is considered in the classification of them. Output parameter is a set of entities, and each entity is a set of attributes. A two-phase classification algorithm is proposed to classifying services with I/O parameters. Services are classified by the entities in the first phase, and classified by attributes of entities in the second phase in this algorithm. In the first phase, the biggest complete graph is separated from a graph in which distance between two random services is recorded, and the number of the biggest complete graph is the number of classes of services. In the seconds phase, a matrix of similarity degree between two random services is used. Find all clusters of services from the matrix according the principle that a service can belong to a cluster and only services which are very similar can belong to the same cluster.The composition of semantic web service of railway is modeled as direct graph in this paper. This model is called general graph of web service composition (GGWSC). Node is service and the edge is the output-input matching relation between two services in the graph. The edge in the graph represents both data flow and control flow. This model can be used in both method of composing services based on output-input matching relation and the hybrid method of services composition.The current methods of composing services based on output-input matching relation between services will be malfunction in certain cases; a new method is proposed to overcome their defects. In this method, search for participated services from candidate services by output-input matching relation, and calculate the Matrix of discovered Services Connection (MWSC), then the biggest composite graph is constructed. Find all smallest composition graphs in this graph and select the one with the optimum QoS (Quality of Service) value. Concepts of share service, a call to a service (CWS) and a list calls to a set of services (LCWS) and tools of matrix of output parameters of web Services (MOPWS) and tree of CWS are used to find out all composition plans from this graph.In process of railway shared information, hybrid composition method is used to construct the model of heterochromatic data statistic. This model is a GGWSC. Use pre-defined method to construct the part known in advance, and use method based on output-input matching relation to construct the unknown part. A multi-level mode of composition request is proposed to collecting information about heterochromatic data statistic.As for the problem of running composite service, a centralizing-distributing schedule mode is proposed to avoid the communication bottleneck caused by centralizing mode and the uncontrolled process caused by distributing mode. In the hybrid mode, running a composite service is still under centralizing control, so how to run a composite service as fast as possible is an important problem. A method to arrange the call order of services which participate in a composition is proposed. Tree of web service composition execution (TWSCE), Order Table of Service Lists (OTSL) and Matrix of Service Lists Connection Value (MSLCV) are used in this method.The research done in this paper provides a certain theoretical support in applying this technology to information sharing in railway enterprise and gives some advices for applying this technology to other enterprises which are similar as railway enterprise.