From the viewpoint of content-based retrieval, the researcher analyzes spatial information of a floor plan, with a focus on the apartment unit floor plan common in Korea. Then the metadata items are extracted in a structured manner. To manage the items efficiently, the researcher develops a data model for spatial information according to the concept of the “Structured Floor Plan”. The main object of content to retrieve is a spatial network that consists of nodes of spaces and their linkages. There are two ways to organize the metadata: the traditional index files and the RDF (Resource Description Framework). While the index files are still efficient with computability, the RDF applies greater options to retrieve, such as fuzzy predicates, semantic predicates, and so on. To exploit the metadata, this research shows several possibilities of query operations that present a set of sample queries about L-DK(Living room – Dining room – Kitchen). Implementation of the prototype system is divided into three parts: 1) a modeling module using Vitruvius; 2) an indexing module using MS SQL Server? 2000 in conjunction XML; and 3) a browsing module using the SpaceScope browser.

The future works may consider XML-based databases and a knowledge based query language, such as RQL/XQL, working on such databases. The more specific domain knowledge is involved, the more practical systems would be. Even in architecture, there may be a diverse range of domain knowledge, such as design, building performance, facility management, energy management, post occupied evaluation, historical research, and so on. Also the issue of interface should be investigated in depth, so that it will be adequate to the needs of the architectural field.