This article introduces a complete storage and retrieval architecture for a database environment for XML documents.
DocBase, a prototype system based on this architecture, uses a flexible storage and indexing technique to allow highly expressive queries without the necessity of mapping documents to other database formats. DocBase is an integration of several techniques that include:
- a formal model called Heterogeneous Nested Relations (HNR),
- a conceptual model XER (Extensible Entity Relationship),
- formal query languages (Document Algebra and Calculus),
- a practical query language (Document SQL or DSQL),
- a visual query formulation method with QBT (Query By Templates), and
- the DocBase query processing architecture.
The paper summarizes experimental and usability analyses to demonstrate its feasibility as a general architecture for native as well as embedded document manipulation methods.