Research On Key Techniques Of Semantic-based Geospatial Information Services Composition

The emergence of geospatial information services provides a new solution for implementingeffectively massive, multi-source, distributed geospatial information to share and interoperation.The contradiction between realistic complex applications and simple function provided by singlegeospatial information service need compose effectively different geospatial information service,and provide new and more powerful value-added service. Geospatial information servicecomposition is a new technology. It has been one of the key scientific problems that geospatialinformation services compose automatically and intelligent. This dissertation researched somekey technology on semantic-based geospatial information services composition, which includedgeospatial information services semantic description model, the visualization and classificationof geospatial information services, geospatial information services chain model and modelvalidation, semantic similarity measurement of geospatial information services andsemantic-based automatic discovery. Aimed at the problems of the existing theory and methods,this dissertation proposed new solution fixed to geospatial information services, the mainresearch and innovation are following:1. The traditional web services description models often have heterogeneous semanticsproblems. This dissertation extended semantic description model of OWL-S according to thefeatures of geospatial information services, and build a new geospatial information semanticdescription model. The proposed model realized accurate, comprehensive and uniformdescription on geospatial information services’ function semantics, data semantics, executionsemantics and QoS semantics.2. This dissertation proposed MDS-based geospatial information services visualizationmethod, and design transformation method from abstract geospatial information services to lowdimensional coordinate. The result of geospatial information services visualization caneffectively guide the selection of the initial number of clusters and cluster centers in K-meansclustering. On this basis, this dissertation proposed the automatic classification method ofgeospatial information services which is based on services dimensionality reduction visualizationand k-means clustering. On one hand, the proposed methods enable people to intuitiveunderstand geospatial information services and structure of the services cluster. On the otherhand, the methods solved the problem that the existing method can’t automatic access to servicesclassified information.3. Due to common web services composition language are’t suitable for intuitive expressing geospatial business processes and are difficult to learn for geospatial users, this dissertationbuilded a geospatial information services chain model based on directed graph, and defined themodel elements, constraints and control mode. This dissertation designed the verification methodof model syntax, structure and semantics, and realized the algorithm and implementation processof the verification method.4. Aimed at these problems that the exsiting web services semantic matchmaking requiredexternal text, can’t quantitative express the similarity and so on, this dissertation proposed thesemantic matchmaking for geospatial information services considering all properties. Theproposed semantic matchmaking method taked into account the information of name, category,description, quality, inputs and output of geospatial information services, integrated the modelbased on semantic reasoning, distance and information content, and is compatible with severalsemantic relation, such as Exact, PlugIn, Subsume and Disjoint.5. This dissertation proposed semantic-based automatic discovery method, which get themost matching cluster through computing the similarity between the Web Service request andeach clustering center in order to reduce the search range for geospatial information servicediscovery. we sort the matching web service according to the semantic similarity considering allproperties between web service request and each matching geospatial information servicebelongs to the most matching cluster. The most matching geospatial information service of webservice request is the most similar web service. The proposed method solved these problems thattraditional automatic discovery methods are lack in semantics and are inefficient.6. Geospatial information service composition system was builded in this dissertation,furthermore, designed and implemented the application case which was that map data qualitychecking. The correctness and validity of proposed models and algorithms were provedaccording to the application case.