Inteligentní systémy pro interaktivní návrh tematických map.

Abstract

In the submitted thesis, an intelligent system for interactive support of thematic map design is designed. The first part describes the decision processes of maps creation. The second part introduces briefly the area of artificial intelligence, describes knowledge and expert systems and their main components. The third part discusses data description standards. It focuses on markup languages since the beginning of their development up to ontology languages forming a basis for semantic descriptions of the real world. The fourth part proceeds to the explanation of knowledge ontologies, to which considerable attention is paid. It mentions processes of acquiring and representing cartographic knowledge, describes basic building components of ontologies – classes, properties and individuals. At the end, it deals with the description of a knowledge ontology development process. In the fifth, main part, the design of an intelligent system is analysed. It begins with the process of construction of a domain knowledge ontology, continues with the issues of domain ontology transformation into Java language interfaces and classes, in the framework of which describes tools suitable for the execution of this transformation. It concerns in detail the transformations of specific parts of the domain ontology – classes, properties and individuals, and draws attention to the areas of possible data loss. Moreover, it deals with the description of several expert systems theoretically applicable to the intelligent system design and focuses on the description of a selected expert system Drools. It concerns the process of processing the ontology, fact base and knowledge base in the Java programming language and demonstrates examples of source codes for the simplest applications. It provides information about cartographic rules and various forms and formalizations of their record. Then it describes possibilities of integration of the expert system Drools with third-party systems, supplemented by practical examples of integration with R statistical software, classification web service and worldwide database of geographical places GeoNames. The last part describes the pilot application CartoExpertGUI. Finally, the results of this doctoral thesis are analysed and evaluated. Appendices to the thesis consist of a hierarchic structure of classes of thematic cartographic ontology, source code for the Java class for work with ontology cardinality, source code simulating user work with the expert system Drools, set of cartographic rules for the intelligent system, source code for the web service Carto Support Service and source code for the Java class for map title analysis.