Semantics-based Fusion for CityGML and 3D LandXML

Abstract

For any developments that require spatial data, often the fusion of diverse spatial datasets is unavoidable. For instance in developing a 3D cadastral database serving various purposes, data may need to be sourced from different spatial datasets such as: building design models in BIM (Building Information Model) format, topographic and built environment information in CityGML, and cadastral legal boundaries in LandXML. Syntactically and systematically this is not difficult provided the formats of datasets involved are open and are XML-based, but the fusion becomes non trivial when semantic heterogeneity occurs between schemas for example between LandXML and CityGML. This paper proposes a semantics-based fusion framework to integrate CityGML and 3D LandXML. The paper adopts the ePlan conceptual model, which is implemented in LandXML in Australia and New Zealand. Compared with the standard LandXML schema, the ePlan model is comprehensive to represent specifically cadastral survey information. The ePlan model is integrated with the Building module of CityGML through the LADM OWL Ontology. In previous research, ISO 19152 Land Administration Domain Model (LADM) has been formalized in Web Ontology Language (OWL). To integrate 3D LandXML and CityGML, this LADM OWL ontology is extended with new concepts: Physical Space Building Unit and Physical Space Utility Network and new relation: hasLegalSpace. Every concept in OWL ontology has a unique URI (Uniform Resource Identifier). The URIs of Physical Space Building Unit and Legal Space Building Unit from the LADM OWL ontology are respectively referenced by CityGML’s ExternalReference element and LandXML’s DocFileRef element. The proposed framework attempts to integrate not only the semantic models inherent in the schemas but also the geometries from CityGML and LandXML. Through this semanticsbased fusion, it is expected that a computer system will be able to do reasoning and inference in the OWL ontology. The computer system will also be able to retrieve the geometries of building unit’s legal space or physical construct, or both, through the ExternalReference and DocFileRef elements. The objective of the framework is to preserve the best of all worlds without changing the existing schemas. Although the framework is a preliminary study and no operational implementation has been done so far, this paper hopes to provide a useful reference in discussing the future directions and harmonization of the schemas.