Integration of Heterogeneous GML Sources

Geography Markup Language (GML), as being an XML Schema meta standard for geospatial data, is becoming widely adopted for both storage and exchange of data between organizations and systems. In particular it is suited for open and seamless exchange of geospatial data on the web. Application schemas are designed based upon a set of GML schemas, allowing for extensions, restrictions and substitution of the base types. Instance documents from different application domains may thus appear heterogeneous, and general-purpose tools are commonly depending upon complex configuration files to handle unknown dialects. We propose a simple dictionary approach enabling integration of differing GML instance documents. The dictionary is built by a cascading process: 1) If all schemas are available, they are analyzed to identify the origin of the specialized elements, or 2) else, if not all schemas are present, or if there are some missing schemas, the instance document(s) are analyzed using certain heuristics based on the base GML element structure, or, 3) if still some element relations are unresolved, the user is asked to assist the system by a manual mapping procedure. We also propose a non-intrusive strategy where we integrate features from differing sources with minimal tampering of the original data. The implemented proof-of-concept framework is far from complete. However, we demonstrate the main features by using the strategy to develop a simple SVG GML browser and apply on to heterogeneous GML data. The work is part of Project OneMap, a long term effort contributing to the fusion of standard web technologies and geographic content, often referred to as the GeoWeb.