Scientific delivery management is critical to building project execution. Ontology-based digital delivery supports structured representation and end-to-end linkage for acceptance and handover. However, interior handover during completion involves numerous elements with tightly coupled spatial relations, while existing knowledge models show limited coverage, unclear responsibility boundaries, and incomplete process chains. A knowledge modeling approach for interior digital delivery under host-component spatial topological constraints is proposed. An IFC-integrated ontology is developed to resolve semantic discontinuities and information inconsistencies. A two-stage extraction algorithm enables batch instance acquisition and synchronizes spatial linkage with BIM-based semantic alignment. Multi-hop graph reasoning, combined with GUID-based deduplication and incremental writing, supports location queries, cross-source fusion, and instance updates. The method was validated on 912 interior deliverables from a furnished residential project in Pudong New Area, Shanghai, where 91.2% precision and 95.1% recall were achieved, with 1,352 instances, 2,132 relations, and 9,196 data properties generated. The method is extensible and can support lifecycle digital delivery and industry digital transformation.