Summary: |
The integration of the GIS and BIM domains attracts unbroken attention in research, because using data from the two worlds in conjunction promises better use of the data, more comprehensive insight and holistic analysis. With the integration of the two domains' data, new use cases become feasible, which are not within the realms of the single domains. Researchers have previously studied various operational modes of integration, from conversion, linking, to interconnected retrieval or generic integration method. Besides a particular integration method, most of the works are also dedicated to a particular pair of schemas, predominantly and most notably the pair of buildingSMART's IFC for digital building models and OGC's CityGML for 3D city models.
Many works highlight the differences between the BIM and GIS domains and contrast them by example of these most prominent schemas of each domain. However, despite standardization efforts, in reality the data and information modelling landscapes of the domains in themselves are not as homogeneous as they appear in the light of such study. There are subdomains within each the BIM and GIS domain with different views and representations of the built environment as well as varying approaches to modelling, ecosystems of tooling for data acquisition and management - and therefor also more than one pertinent standard in each domain. Thus, integrating pairs of schemas is not sufficient, we must extend the horizon by looking at how to integrate more than two schemas.
In this paper we present a study on a use case driven approach to integrating multiple schemas, namely IFC, CityGML, IndoorGML and OpenStreetMap. Instead of a generic integration, at this point, we look only at a specific operational mode of integration which is conversion from building models into data sets for city maps and navigation services. This is guided by the practical aim of a feasibility study carried out recently: how to enhance and extend city maps and navigation models with indoor information originating from planning processes.
Simply lumping together isolated pairwise considerations can certainly cover the ground of the various facets of both domains, but not assure consistency overall. It seems likely that compared to pairwise model integration, with multiple models, complexity increases substantially and even subtle inconsistencies between two models potentiate to significant impact. The study was conducted as an experiment to get a first heuristic understanding of the increase in complexity, to identify and try-out different viable methods of forming sub sets or super sets of the models, or complementary structures to the models. These identified model pieces are then to assembled in an intermediate model in order to facilitate the conversion via a mediating step.
The full paper will describe the identified mediating model elements, demonstrate the details of the intermediate model and how it bridges the partially contrasting modelling paradigms of IFC and the target formats. We also show the implementation of the conversion from the digital building model to the intermediate model and simultaneous identification of model requirements. The proof-of-concept implementation has been tested with a set of sample data and we show the results. We discuss the limitations of the study, challenges and opportunities to develop a more formal approach to multiple schema integration which also takes into account the different operational integration modes beyond conversion. |