On the Complexity of Standard and Specialized DTD Parsing

We investigate an extension of standard DTDs named s-xschemes yielding more expressive power. We show that s-xschemes which are also known as specialized DTDs have more convenient complexity properties than context-free languages. We invent the notions of proper-ness and 1-unambiguity for a close characterization of s-xschemes, whereby \1-unambiguity" is a restriction concerning expressive power while \properness" is not. With these conditions we show an eecient algorithm for parsing in linear time and space. We propose a concrete syntax and provide an implementation inside a parsing system.