A Cost-Based Approach for Converting Relational Schemas to XML

Converting relational database schemas to XML schemas is important as applications originally written to manipulate relational data are converted to XML-enabled applications. Most research has focused on the challenging problem of efficiently extracting relational data through XML views. However, another important challenge is determining good XML schemas from existing relational schemas, so that applications may be migrated to use XML. In this paper, we propose a cost measure based on the space efficiency of the XML encoding as a metric for measuring the XML schema’s desirability. Using this cost metric, we define an algorithm that returns the optimal XML encoding exhibiting no redundancy. Further, the algorithm can calculate the most space efficient encoding with respect to user constraints on the XML schema, which has previously not been possible without forcing the user to specify a complete mapping. The result is a fully automatic system for migrating relational schemas to XML schemas while respecting conversion constraints.