Personal tools
You are here: Home Events Abstract Archives 2003 Principles of typed XML processing

Principles of typed XML processing

Haruo Hosoya Research Institute for Mathematical Sciences Kyoto University 4pm Tuesday 11 March 2003 Room 2511, JCMB, King's Buildings

XML is a standard exchange format for tree structure data and quickly became popular recently. XML's main strength is its ability to give structural constraints on data by using types.

In this talk, I overview the series of work done in the XDuce project for designing a statically typed language for XML processing. The key technique here is to use tree language theory as the basis. Specifically, we adopt tree regular expressions as types, extend these to a pattern matching facility, and perform typechecking by operations on them. Necessary algorithmic problems are then solved by converting types to finite tree automata. In particular, we can thus use the language inclusion between tree regular expressions as subtyping, which is literally the most powerful since it allows all subtype relations that are semantically sound. We also use boolean operations such as intersection and difference for analyses on pattern matches. The resulting type system is simple, elegant, and expressive and thereby should be easy for the user to understand. I also mention our recent efforts for incorporating the attribute typing mechanism of RELAX NG to our type system.

This work is an outcome from my collaboration with Benjamin Pierce, Jerome Vouillon, and Makoto Murata.

Document Actions