Standardisation, archiving, and digital access of spatial data pertaining to built-up environments is an area acquiring increasing attention amongst several interest groups: policy makers, designers and planners, civil engineers, infrastructure management and public service personnel, building users. Initiatives such as the Building Information Model (BIM), Industry Founda- tion Classes (IFC), and CityGML are creating the information-theoretic backbone that guides the crucial aspects of quality, exchange, and interoperability of spatial data at the environmental and urban scale. However, due to the inherent scale, com- plexity, and detailed geometric character of building information data, extracting useful semantic and qualitative knowledge for accomplishing high-level analytical tasks is still an extremely complex and error prone process involving data intensive computing. We propose a uniform spatial data access middleware that can provide a combination of high-level, multi-modal, semantic, and quantitative-qualitative spatial data access and analytical capability. We present the core computational capabil- ities for the proposed middleware and present an overview of the high-level spatial model and its compliance with the industry standard IFC. A key theoretical contribution is a framework for investigating the computational complexity of deriving spatial artefacts within the context of building informatics. Additionally, we empirically investigate the feasibility and practicality of the derivation of spatial artefacts by conducting experiments on seven industry-scale IFC models. The experiment results show that, despite having non-linear polynomial increase with respect to time, deriving spatial artefacts is practical with large designs.