Diplomarbeit Susanne Reinwarth

Untersuchungen zur Verwendung des Rule Interchange Formats (RIF) für die dienstbasierte Transformation von Geodaten

Abstract
The harmonization of spatial data sets is one of the major challenges for the development of spatial data infrastructures (SDI). Therefore, schema transformations are applied to convert the original data into a unified data schema using specified transformation rules, called the schema mapping. For the implementation of INSPIRE (Infrastructure for Spatial Information in Europe) a web-based schema transformation service is proposed, which performs the schema transformation based on a generic schema mapping described in the Rule Interchange Format (RIF). RIF is a standard of the World Wide Web Consortium (W3C), which was primarily designed for the exchange of rules between different rule systems.This thesis has examined the suitability of RIF for the service-based transformation of spatial data in the context of INSPIRE, with a particular focus on the coding of mapping rules. As test case, a given schema mapping between German cadastral data (AFIS-ALKIS-ATKIS) and INSPIRE Cadastral Parcels has been expressed in RIF rules. To achieve this, the RIF export of the deployed schema mapping software HALE (HUMBOLDT Alignment Editor) had to be extended. The gained experiences on the description of a concrete schema mapping with RIF and the integration of RIF into an existing schema mapping software are described in this work. Furthermore, the performance of RIF has been extensively tested using six schema transformation levels. These levels include different types of transformations with varying complexity and special regard to spatial functions, which may be relevant for schema transformations for INSPIRE. The suitability of RIF as an implementation independent description language for schema mappings has been discussed from various points of view and was finally confirmed. The main reasons were the expressiveness of the language, its full support of all six schema transformation levels and the test case as well as the uncomplicated extensibility to spatial functionality. However, for schema mappings, the procedural character of RIF proved to be less intuitive than a declarative mapping language. In addition, the complexity of the language was in some points more extensive. Therefore, the integration of RIF in existing software was found to be complex, especially if the particular schema mapping software or transformation tool is based on a declarative mapping language.