Research On Geographical Information Service Discovery Methods

Geographical information service effectively promotes the sharing and functional reuse ofgeographical information. At present, more and more enterprises or organizations open their owndata or software functions as geographical information services to be used by people. However,following geographical information services on the web are increasing, people want to getsatisfied geographical information services become more and more difficult. Therefore, highlyefficient geographical information service discovery methods are urgently needed to help peoplefind and select required services. So this dissertation imports the techniques of information query,semantic web, data minning, and multiple attributes comprehensive evaluation to discuss thegeographical information service discovery methods. The methods of geographical informationservice discovery based on description and simple semantic, geographical information servicediscovery based on ontology semantic and rule support, geographical information service basedon classification and clustering, geographical information service discovery based on QoS arediscussed in detail. The main achievement and innovation are described as follows.1. The backgrounds, significances, and related theories of geographical information servicediscovery are anlalyzed. The geographical information system service-oriented changing, andexisted problems in the process of geographical information service sharing are analysed. Thenthe significances of geographical information service discovery are presented. Web service,semantic web, and web service QoS evaluation are confirmed as the three basic techniques ofgeographical information service discovery. The definition of geographical information servicediscovery is given, and the framework and methods are discussed. Finally, the evaluationmethods of geographical information service discovery are given.2. On now web environment, geographical information service discovery method based onkeywords cannot get satisfying effect. The dissertation imports the information query techniqueand WordNet lexical semantic technique to realize the geographical information servicediscovery methods based on description and simple semantic. The geographical informationservice discovery method based description is realized by service name matching with editdistance and service description matching with vector space model. The method of simplesemantic based geographical information service discovery is realized by constructing virtualdocuments and imporing WordNet lexical semantic. This method achieves the operation levelservice matching, and can support the simple semantic of synonymy, hypernym and hyponym.3. The geographical information service discovery methods based on ontology semantic andrule are discussed to sovle the problem of service finding on semantic web entironment. Aiming the problem, the geographical information ontology building rules, building methods, buildingtools, logic composition, and integration methods are discussed. Geographical informationservice semantics are confirmed by containing data or information semantic, function oroperation semantic, execution semantic, and QoS semantic. Then OWL-S is used to semanticlydescribe geographical information service. Weighted semantic distance and Wu-Palmer method arecombined to improve the calculating method of ontology concept semantic similarity, and thenservice interfaces dependent relations are introduced to promot the geographical informationservice input/output(IO) matching method which supports interface multi-condition. After that,geographical information service precondition/effect (PE) matching method is promoted.4. Geographical information services become disordered and irregular because of bigamounts and multifarious classes, which leads to the low service discovery efficiency. Aimingthe problem, the dissertation promots using service classification and clustering for geographicalinformation service discovery. Therefor, the geographical information service classificationcriterion is discussed for the preparation of machine classification label. Na ve BayesClassification algorithm and k-Nearest Neighbor algorithm are used to do geographical informationservice auto-classification, and are examined by experiments. Then geographical informationservice classification matching method is presented. Geographical information service k-Meansclustering algorithm based on service characters is promoted, and some problems are solved,such as selecting initial cluster centers, re-confirming cluster centers on the process, expressingservice clusers after clustering, and dealing with exceptional services. Then geographicalinformation service cluster matching method is presented. Finally, experiments are practised toexamine the application effects of service classification and clustering.5. In order to filter the geographical information services from non-function attributes, thegeographical information service discovery methods based on QoS are discussed. Geographicalinformation service publication and discovery framework based on QoS is promoted, whichespecially imports QoS broker. After analyzing the geographical information service QoSattributes’ characters, QoS model based on service classification which can extend is promoted.Geographical information service QoS attributes order-relation is introduced to calculate thepowers of QoS attributes. Then two geographical information service QoS evaluation methodsare discussed, they are simple additive weighted method and fuzzy comprehensive evaluationmethod. Geographical information service matching methods of QoS preference model based onorder-relation and weight are discussed. Finally, experiments are practised to examine theapplication effects of QoS evaluation and discovery method based on QoS.