QL '98 Position Paper

Shekhar Shastri, Redwood Risk Management, USA


XML brings three promises that are of interest to us:

Our requirements from a query language for XML are:

Data Model: The data model underlying XML should permit loosely structured data as well as well-structured. [Abiteboul et al] [Michael Rys].

Data Transformation: Powerful data transformation primitives must exist to allow for selection, classification, aggregation and extraction. [Deutsch, Fernandez, Florescu, Levy, Suciu]

Customization: Multiple views of the data based upon user specified preferences must be possible. User-specific dynamic content generation should be simple to accomplish. A naive user must be able to dynamically create a view on the fly, even if the corresponding stylesheet is not available.

Ease of  Use: The widespread adoption of the XML Query standard would depend much on the ease with which it can be used with little no training. Thus, there good reasons to use a URI based syntax for the query language. Yet, limiting the query syntax to URIs will severely limit the expressive power of the query language. Thus, we feel that there is a need for multiple syntax for the query language addressing both the simple URI kind of syntax as well as a declarative syntax that builds upon other database query language research and standards.[Michael Rys] [William Shea, Paul Kanevsky, Ramesh Lekshmynarayanan]