A Framework for a Linked Data-based Heritage BIM

Collaborative built heritage projects require that a large variety of stakeholders, often using different tools and datasets, collaborate efficiently to conserve a valuable heritage building. Despite the opportunities of a thorough digitization, the general adoption and exchange of structured data during such projects is currently lower compared to other sectors of the construction domain focusing on new buildings. Besides conventional Building Information Modeling (BIM) applications and related exchange formats such as IFC (Industry Foundation Classes), researchers have been studying the potential of standardized Semantic Web Technologies (SWT), including Linked Data graphs, to improve the modeling of structured building data and its exchange over the Web. In this thesis, it is argued that the (Heritage) BIM concept can be implemented using SWT, resulting in a framework for a Linked Data-based Heritage BIM (LD-based HBIM), instead of relying on conventional BIM technology alone. The application of SWT requires the availability of a shared data structure, documented in so-called web ontologies, to obtain rich and interoperable building descriptions. A network of modular ontologies is proposed in contrast to a few large and complex ontologies. In many cases, terminology documented in these modular ontologies is expected to be applicable beyond the heritage domain, e.g. for the description of new buildings or other types of constructions. Besides ontologies, workflows and proof-of-concept applications are considered valuable to demonstrate how the LD-based HBIM framework can be applied in practice. First, the characteristics and application of SWT in built heritage projects, besides other data technologies, are studied through a literature review. Second, current workflows in the Flemish built heritage sector and opportunities for digitization are collected through semi-structured interviews with stakeholders. The results of this stakeholder analysis are used to derive five thematic sets of requirements - including overall needs and requirements related to the description of buildings in general, geometry, damage and construction dataset metadata - as the benchmark for the development of the LD-based HBIM framework. Using the benchmark requirements, an ontology network is prepared consisting of existing, transformed and, wherever necessary, new ontologies. In the proposed ontology network, a division is made between (1) concise core ontologies for the description of buildings and related aspects, (2) extending ontologies and taxonomies for properties, geometry metadata and classification, and (3) a separate set of terminology for construction dataset metadata. According to best practices, existing ontologies such as BOT (building topology) and DCAT (general dataset metadata) are reused wherever possible. The OMG (geometry linking), OPM (properties versioning) and DOT (damage linking and topology) ontologies are newly developed together with other researchers. The new CTO (construction tasks), ConTax (taxonomies management) and CDC (construction dataset context) ontologies are a direct result of this research. Existing taxonomies related to the classification of construction components, spatial building zones and materials are transformed to make them suitable for application with the core ontologies. In addition, new taxonomies are prepared for properties (CP and BHP), geometry formats (FOG) and the classification of damages (MDCS-O and MWV-D) and tasks (MVW-T). Finally, the new GOM ontology can be combined with OMG and FOG to facilitate the reuse of geometry descriptions in any existing geometry format. Different workflows, accompanied by existing and newly developed demo applications, for the authoring of building datasets are identified and analyzed. The modeling of construction dataset metadata and the exchange of datasets are documented in a higher-level workflow. The developed ontology network is applied in five example cases, inspired by two built heritage projects in Ghent, to demonstrate the new functionalities of the developed LD-based HBIM framework. The developed ontology network proves valuable for the description of unique and potentially complex heritage buildings thanks to its modularity, concise core ontologies and extendable taxonomies for classification and properties. A large part of the developed ontologies can be reused to describe other types of constructions. The thesis demonstrates how technical and semantic interoperability can be established by applying the LD-based HBIM framework while creating, exchanging and reusing construction datasets.

Jaar van publicatie:2021

Toegankelijkheid:Open