The emergence of the digital twin concept can potentially change the way people manage built assets thoroughly. This is because the semantics-based model and linked data approach behind the digital twin, as the successor of classical BIM, provide strong capability in integrating data from fragmented and heterogeneous sources and thus enables better-informed decision-making. Taking buildings as the case, this paper demonstrates the ontology-based Information Management Framework and elaborates the process to integrate data through a common data model. Specifically, the Foundation Data Model (FDM) representing the operation of buildings and embedded systems is developed and two patterns of integration architecture are compared. To conceptualise all the essential entities and relationships, the building topology ontology and BRICK ontology are reused and merged to serve as a feasible FDM. According to the characteristic of asset management services that digital twin supports, two integration architectures are compared, including the data warehouse approach and mediator approach. A case study is presented to elaborate the implementation of these two approaches and their applicabilities. This work sets out the standardised and modularised paradigms for discovering, fetching, and integrating data from disparate sources with different data curation manners.