Efficient XSLT processing in relational database system

Efficient processing of XQuery, XPath and SQL/XML on XML documents stored and managed in RDBMS has been widely studied. However, much less of such type of work has been done for efficient XSLT processing of XML documents stored and managed by the database. This is partially due to the observation that the rule based template driven XSLT execution model does not fit nicely with the traditional declarative query language processing model which leverages index probing and iterator based pull mode that can be scaled to handle large size data. In this paper, we share our experience of efficient processing of XSLT in Oracle XML DB. We present the technique of processing XSLT efficiently in database by rewriting XSLT stylesheets into highly efficient XQuery through partially evaluating XSLT over the XML documents structural information. Consequently, we can leverage all the work done for efficient XQuery/XPath processing in database to achieve combined optimisations of XSLT with XQuery/XPath and SQL/XML in Oracle XMLDB. This effectively makes XSLT processing scale to large size XML documents using classical declarative query processing techniques in DBMS.