We present a methodology for decomposing documents by content and integrating these into a rich information structure. This implies both expanding the document structure, replacing document entities by detailed substructures, and augmenting the structureis relatedness with content information. This paper focuses on some of the representational issues involved in the process of interpreting, breaking up, and relating documents. We describe a prototype application as a tool for building up, storing, and presenting architectural analyses in an educational setting implemented using XML, discuss a similar prototype application to be implemented using sorts, and compare these two different methodologies.