Abstract
Existing schema languages can lead to overdesign. They offer more choices than are necessary for purely descriptive (as contrasted with prescriptive) situations. A potential solution is to design based on "patterns" from real DTDs. Using three example situations, alternatives, repeatable homogeneous elements, and mixed content models, we derived a group of patterns sufficient to express all required structures in a descriptive environment. To provide a meaningful example, we propose a new instance-based schema language, DTD--, that derives schemas from tagged sample instances according to the patterns. Since there are few patterns, every document can be represented by a simple grammar where grammar rules can be directly inferred from the document, without any ambiguity.
Users
Please
log in to take part in the discussion (add own reviews or comments).